顔写真

イヌイ ケンタロウ
乾 健太郎
Kentaro Inui
所属
言語AI研究センター
職名
教授
学位
  • 博士(工学)(東京工業大学)

  • 修士(工学)(東京工業大学)

経歴 9

  • 2023年9月 ~ 継続中
    Mohamed bin Zayed University of Artificial Intelligence Visiting Professor

  • 2021年9月 ~ 継続中
    特定国立研究開発法人理化学研究所 革新知能統合研究センター 自然言語理解チーム リーダー

  • 2020年4月 ~ 継続中
    タフ・サイバーフィジカルAI研究センター 副センター長・AI研究部門長

  • 2016年4月 ~ 継続中
    理化学研究所 革新知能統合研究センター 自然言語理解チームリーダー

  • 2010年4月 ~ 継続中
    東北大学 教授

  • 2007年10月 ~ 2010年3月
    奈良先端科学技術大学院大学情報科学研究科 准教授

  • 2002年4月 ~ 2007年9月
    奈良先端科学技術大学院大学 情報科学研究科 助教授

  • 1998年4月 ~ 2002年3月
    九州工業大学情報工学部 助教授

  • 1995年4月 ~ 1998年3月
    東京工業大学大学院情報理工学研究科 助手

︎全件表示 ︎最初の5件までを表示

学歴 2

  • 東京工業大学 情報理工学研究科 計算工学専攻

    ~ 1995年3月

  • 東京工業大学 工学部 情報工学科

    ~ 1990年3月

委員歴 43

  • 言語処理学会 副会長

    2020年4月 ~ 継続中

  • 日本学術会議 連携会員

    2017年10月 ~ 継続中

  • ファクトチェック・イニシアティブ 理事

    2017年10月 ~ 継続中

  • 言語処理学会 理事

    2016年6月 ~ 継続中

  • ACL-IJCNLP 2021 Workshop Co-chair

    ~ 2021年

  • COLING 2020 Area Co-chair

    ~ 2020年12月

  • ACL 2020 Area Co-chair

    ~ 2020年7月

  • EMNLP 2019 General Chair

    2018年7月 ~ 2019年11月

  • 言語処理学会 会誌「自然言語処理」 編集長

    2018年4月 ~ 2019年3月

  • 言語処理学会 論文誌副編集長

    2016年6月 ~ 2018年3月

  • 情報処理学会 自然言語処理研究会主査

    2014年4月 ~ 2018年3月

  • COLING2016 Program Committee Area Co-chair

    2016年5月 ~ 2016年9月

  • COLING2016 Program Committee Area Co-chair

    2016年5月 ~ 2016年9月

  • 言語処理学会 第22回年次大会 実行委員長

    2015年4月 ~ 2016年3月

  • 言語処理学会 第22回年次大会 実行委員長

    2015年4月 ~ 2016年3月

  • EMNLP2015 Program Committee Area Co-chair

    2015年4月 ~ 2015年8月

  • EMNLP2015 Program Committee Area Co-chair

    2015年4月 ~ 2015年8月

  • 情報処理学会 論文誌編集委員長

    2014年6月 ~ 2015年5月

  • 情報処理学会 論文誌編集委員長

    2014年6月 ~ 2015年5月

  • 情報処理学会 理事

    2013年6月 ~ 2015年5月

  • 情報処理学会 理事

    2013年6月 ~ 2015年5月

  • COLING2014 Program Committee Area Co-chair

    2014年1月 ~ 2014年8月

  • COLING2014 Program Committee Area Co-chair

    2014年1月 ~ 2014年8月

  • 情報処理学会 論文誌副編集委員長

    2013年6月 ~ 2014年6月

  • 情報処理学会 論文誌副編集委員長

    2013年6月 ~ 2014年6月

  • Association for Natural Language Processing Board Member

    2010年6月 ~ 2014年5月

  • 言語処理学会 理事

    2010年6月 ~ 2014年5月

  • Association for Natural Language Processing 編集委員

    2009年10月 ~ 2013年9月

  • 言語処理学会 編集委員

    2009年10月 ~ 2013年9月

  • Association for Computational Linguistics Editorial Board Member

    2008年12月 ~ 2012年11月

  • Association for Computational Linguistics 編集委員

    2008年12月 ~ 2012年11月

  • ACL2012 Program Committee Faculty Advisor for Student Research Workshop

    2012年2月 ~ 2012年8月

  • ACL2012 Program Committee Faculty Advisor for Student Research Workshop

    2012年2月 ~ 2012年8月

  • 情報処理学会 論文誌査読委員

    2006年6月 ~ 2012年5月

  • 情報処理学会 論文誌査読委員

    2006年6月 ~ 2012年5月

  • IJCNLP2011 Program Committee Tutorial Co-chair

    2010年8月 ~ 2011年11月

  • IJCNLP2011 Program Committee チュートリアル共同委員長

    2010年8月 ~ 2011年11月

  • The 17th Annual Meeting of Natural Language Processing PC Chair

    2010年8月 ~ 2011年8月

  • 言語処理学会第17回年次大会プログラム委員会 プログラム委員長

    2010年8月 ~ 2011年8月

  • ACL-HLT2011 Program Committee Area Chair for Discourse and Pragmatics

    2010年9月 ~ 2011年3月

  • ACL-HLT2011 Program Committee 談話・語用論分野委員長

    2010年9月 ~ 2011年3月

  • 社会的知能発生学研究会 幹事

    2008年10月 ~

  • 社会的知能発生学研究会 幹事

    2008年10月 ~

︎全件表示 ︎最初の5件までを表示

所属学協会 5

  • 人工知能学会

  • 情報処理学会

  • 言語処理学会

  • American Association for Artificial Intelligence

  • Association for Computational Linguistics

研究キーワード 11

  • natural language inference

  • 自然言語推論

  • 自然言語対話

  • 意味処理

  • 知識獲得

  • Web情報処理. 情報編集

  • 知識処理

  • 自然言語理解

  • コミュニケーション科学

  • 計算言語学

  • 自然言語処理

研究分野 1

  • 情報通信 / 知能情報学 / コミュニケーション科学

受賞 97

  1. NLP若手の会第18回シンポジウム Helpfeel賞

    2023年8月 日本語学習のための形態意味中心の動詞活用

  2. The 17th International Workshop on Semantic Evaluation (SemEval-2023) Nominated for Best System Paper

    2023年7月 SemEval TohokuNLP at SemEval-2023 Task 5: Clickbait Spoiling via Simple Seq2seq Generation and Ensembling

  3. 言語処理学会第29回年次大会(NLP2023)委員特別賞

    2023年3月 言語処理学会 What can Short Answer Scoring Models Learn from Cross-prompt Training Data?

  4. 言語処理学会第29回年次大会(NLP2023)委員特別賞

    2023年3月 言語処理学会 言語モデルの学習における知識ニューロンの形成過程について

  5. 言語処理学会第29回年次大会(NLP2023)委員特別賞

    2023年3月 言語処理学会 ニューラル数式ソルバーにおける途中結果の追跡と操作

  6. 言語処理学会第29回年次大会(NLP2023)委員特別賞

    2023年3月 言語処理学会 日本語話者の項省略判断に関するアノテーションとモデリング

  7. 言語処理学会第29回年次大会(NLP2023)委員特別賞

    2023年3月 言語処理学会 計算資源が限られた複数組織での出力選択による協働の検討

  8. 言語処理学会第29回年次大会(NLP2023)日本電気賞

    2023年3月 言語処理学会 計算資源が限られた複数組織での出力選択による協働の検討

  9. 言語処理学会第29回年次大会(NLP2023)言語資源賞

    2023年3月 言語処理学会 日本語日常対話コーパスの構築

  10. 言語処理学会第29回年次大会(NLP2023)優秀賞

    2023年3月 言語処理学会 Cross-stitching Text and Knowledge Graph Encoders for Distantly Supervised Relation Extraction

  11. 言語処理学会2022年度最優秀論文賞

    2023年3月 言語処理学会 負例を厳選した対話応答選択による対話応答生成システムの評価

  12. AACL-IJCNLP 2022 Student Research Workshop (SRW) Best Paper Award

    2022年12月 AACL Empirical Investigation of Neural Symbolic Reasoning

  13. システムライブコンペティション5 優秀賞

    2022年12月 人工知能学会 aoba_v3 bot: 多様な応答生成モデルとルールベースを統合したマルチモーダル雑談対話システム

  14. The 23rd International Conference on Artificial Intelligence in Education (AIED2023) Best Paper Nominee

    2022年7月 AIED Balancing Cost and Quality: An Exploration of Human-in-the-loop Frameworks for Automated Short Answer Scoring

  15. 科学技術分野の文部科学大臣表彰 科学技術賞

    2022年4月 文部科学省 知識と推論による言語情報編 集の計算モデルの研究

  16. 言語処理学会2021年度論文賞

    2022年3月 言語処理学会 訓練事例の影響の軽量な推定

  17. 言語処理学会第28回年次大会 委員特別賞

    2022年3月 言語処理学会 動画タイトルを用いたサムネイル画像の自動選択手法の提案

  18. 言語処理学会第28回年次大会 委員特別賞

    2022年3月 言語処理学会 Transformerにおけるフィードフォワードネットの作用

  19. 言語処理学会第28回年次大会 スポンサー賞(富士通賞)

    2022年3月 言語処理学会 Transformerモデルのニューロンには局所的に概念についての知識がエンコードされている

  20. 言語処理学会第28回年次大会 スポンサー賞(NEC賞)

    2022年3月 言語処理学会 四則演算を用いた Transformer の再帰的構造把握能力の調査

  21. 言語処理学会第28回年次大会 優秀賞

    2022年3月 言語処理学会 論述リビジョンのためのメタ評価基盤

  22. 第93回 人工知能学会 言語・音声理解と対話処理研究会(SLUD)第12回対話システムシンポジウム 対話システム ライブコンペティション4 優秀賞

    2021年11月 人工知能学会 aoba_v2 bot: 多様な応答生成モジュールを統合した雑談対話システム

  23. 情報処理学会 第249回自然言語処理研究会 優秀研究賞

    2021年7月 情報処理学会 事例ベース推論を行うニューラルモデルの説明性とハブ現象の関係

  24. 言語処理学会第27回年次大会 サイバーエージェント賞

    2021年3月 言語処理学会 指定語句を確実に含む見出し生成

  25. 言語処理学会第27回年次大会 委員特別賞

    2021年3月 言語処理学会 指定語句を確実に含む見出し生成

  26. 言語処理学会第27回年次大会 委員特別賞

    2021年3月 言語処理学会 項目採点技術に基づいた和文英訳答案の自動採点

  27. 言語処理学会2020年度最優秀論文賞

    2021年3月 言語処理学会 論述構造解析におけるスパン分散表現

  28. 言語処理学会第27回年次大会 デンソーアイティーラボラトリ賞

    2021年3月 言語処理学会 単語埋め込みの決定的縮約

  29. 言語処理学会第27回年次大会 委員特別賞

    2021年3月 言語処理学会 予測の正確な言語モデルがヒトらしいとは限らない.

  30. 言語処理学会第27回年次大会 委員特別賞

    2021年3月 言語処理学会 対話システムの矛盾応答の生成に対する脆弱性の分析

  31. 言語処理学会第27回年次大会 委員特別賞

    2021年3月 言語処理学会 オープンドメイン質問応答における解答可能性判別の役割

  32. 言語処理学会第27回年次大会 優秀賞

    2021年3月 言語処理学会 単語埋め込みの決定的縮約

  33. 第3回対話システムライブコンペティション 優秀賞

    2020年11月 ILYS aoba bot: 大規模ニューラル応答生成モデルとルールベースを統合した雑談対話システム

  34. 人工知能学会 全国大会優秀賞

    2020年11月 人工知能学会 ニューラルネットワークによる自然言語推論の可能性

  35. 人工知能学会 全国大会優秀賞

    2020年11月 人工知能学会 単語埋め込みのノルムと⽅向ベクトルを区別した⽂間最適輸送コスト

  36. 言語処理学会第26回年次大会 最優秀賞

    2020年3月 言語処理学会 ベクトル⻑に基づく自己注意機構の解析

  37. 言語処理学会第26回年次大会 最優秀賞

    2020年3月 言語処理学会 超球面上での最適輸送に基づく文類似性尺度

  38. 言語処理学会第26回年次大会 優秀賞

    2020年3月 言語処理学会 テキストを通して世界を見る:機械読解における常識的推論のための画像説明文の評価

  39. 言語処理学会第26回年次大会 優秀賞

    2020年3月 言語処理学会 大規模疑似データを用いた高性能文法誤り訂正モデルの構築

  40. 言語処理学会第26回年次大会 言語資源賞

    2020年3月 言語処理学会 機械読解システムの推論過程のベンチマークの構築

  41. 言語処理学会第26回年次大会 言語資源賞

    2020年3月 言語処理学会 小規模リソースにおける生成型要約のためのスタイル転移

  42. 人工知能学会 音声・言語理解と対話処理研究会(SLUD)第87回研究会 第10回対話システムシンポジウム 若手萌芽ポスター賞

    2019年12月 人工知能学会 負例を厳選した対話応答選択テストセット構築の試みと分析

  43. 情報処理学会 自然言語処理研究会 優秀研究賞

    2019年8月 情報処理学会 クイズ解答タスクにおける大規模ラベルなしコーパスの利用: 言語モデルとデータ拡張

  44. NLP若手の会第14回シンポジウム 奨励賞(三田雅人)

    2019年8月 NLP若手の会 文法誤り訂正を拡張した新タスクの提案

  45. NLP若手の会第14回シンポジウム 奨励賞(横井祥)

    2019年8月 NLP若手の会 文ベクトルの最適輸送に基づく類似性尺度

  46. 言語処理学会第25回年次大会 言語資源賞

    2019年3月 言語処理学会 解説文生成研究のためのライティング技術解説付き学習者コーパス

  47. 言語処理学会第25回年次大会 優秀賞

    2019年3月 言語処理学会 ExpertとImitatorの混合ネットワークによる大規模半教師あり学習

  48. 全国大会優秀賞(横井祥)

    2018年11月 人工知能学会 カーネル法に基づく疎な言語表現のための高速計算可能な共起尺度

  49. 全国大会優秀賞(赤間怜奈)

    2018年11月 人工知能学会 スタイルの類似性を捉えた単語ベクトルの教師なし学習

  50. 第1回対話システムライブコンペティション 優秀賞

    2018年11月 Zunkobot: 複数の知識モジュールを統合した雑談対話システム

  51. NLP若手の会第13回シンポジウム 奨励賞(白井穂乃)

    2018年8月 人工知能学会 情報科学論文からの技術の利点・欠点のマイニングに向けて

  52. NLP若手の会第13回シンポジウム 奨励賞(佐々木翔大)

    2018年8月 NLP若手の会 サブワードに基づく単語ベクトルの再構築

  53. 人工知能学会2017年度論文賞

    2018年6月 人工知能学会 関係パタンの分散表現の計算 (人工知能学会誌, Vol.32, No.4, pp. D-G96_1-11, 2017)

  54. 言語処理学会第24回年次大会 優秀賞

    2018年3月 言語処理学会 ニューラルヘッドライン生成における誤生成問題の改善

  55. 言語処理学会第24回年次大会 若手奨励賞(阿部香央莉)

    2018年3月 言語処理学会 ニューラルネットを用いた多方言の翻訳と類型分析

  56. 言語処理学会第24回年次大会 若手奨励賞(鈴木正敏)

    2018年3月 言語処理学会 読解による解答可能性を付与した質問応答データセットの構築

  57. 言語処理学会第24回年次大会 若手奨励賞(佐藤祥多)

    2018年3月 言語処理学会 因果関係に基づくデータサンプリングを利用した雑談応答学習

  58. 言語処理学会第24回年次大会 若手奨励賞(佐々木翔大)

    2018年3月 言語処理学会 言語横断的情報検索の大規模データセットとパラメータ共有モデル

  59. 情報処理学会第166回データベースシステム研究会ならびに電子情報通信学会データ工学研究会(食メディア研究会協賛) 学生奨励賞

    2017年12月 食メディア研究会 雛形化による手続き的知識の汎化

  60. 人工知能学会 音声・言語理解と対話処理研究会(SLUD)第81回研究会 第8回対話システムシンポジウム 若手奨励賞

    2017年10月13日 人工知能学会 発話の語の文体ベクトルの半教師あり学習

  61. NLP若手の会第12回シンポジウム 奨励賞

    2017年9月6日 NLP若手の会 カーネル密度推定に基づく関係予測

  62. NLP若手の会第12回シンポジウム 奨励賞

    2017年9月6日 NLP若手の会 発話スタイル空間の教師なし学習およびスタイル制御可能な対話システムの実現

  63. 情報処理学会 自然言語処理研究会 学生奨励賞

    2017年5月16日 情報処理学会 自然言語処理研究会

  64. EACL2017 Outstanding Paper Award

    2017年4月7日 European Chapter of the Association for Computational Linguistics

  65. 言語処理学会第23回年次大会 優秀賞

    2017年3月16日 言語処理学会

  66. PACLIC-30 Best Paper Honorable Mention

    2016年10月29日 The 30th Pacific Asia Conference on Language, Information and Computation

  67. MIRU2016 Outstanding Paper Award

    2016年8月 MIRU2016 Learning to Describe Product Images from Noisy Online Data

  68. 2016 IBM Faculty Award

    2016年6月 IBM

  69. 言語処理学会第22回年次大会 優秀賞

    2016年3月10日 言語処理学会

  70. 言語処理学会第22回年次大会 最優秀賞

    2016年3月10日 言語処理学会

  71. 情報処理学会 自然言語処理研究会 優秀研究賞

    2016年1月22日 情報処理学会

  72. PACLIC-29 Best Paper Award (Computation)

    2015年11月1日 PACLIC

  73. ドコモ・モバイル・サイエンス賞 先端技術部門優秀賞

    2015年9月28日 ドコモ

  74. 情報処理学会 自然言語処理研究会 優秀研究賞

    2015年5月26日 情報処理学会

  75. 言語処理学会第21回年次大会 最優秀賞

    2015年3月19日 言語処理学会

  76. 言語処理学会2014年度最優秀論文賞

    2015年3月19日 言語処理学会

  77. 日本学術振興会 科学研究費審査委員表彰

    2014年10月31日 日本学術振興会

  78. AMT2014 Best Paper Award

    2014年8月13日 AMT

  79. 言語処理学会 20周年記念論文賞

    2014年7月11日 言語処理学会

  80. 言語処理学会第20回年次大会 最優秀賞

    2014年3月20日 言語処理学会

  81. 言語処理学会第20回年次大会 優秀賞

    2014年3月20日 言語処理学会

  82. 言語処理学会 2013年度論文賞

    2014年3月19日 言語処理学会

  83. 言語処理学会第19回年次大会 優秀賞,2013年8月21日

    2013年11月16日 言語処理学会

  84. 東北大学 工学研究科長教育賞

    2013年3月22日 東北大学

  85. 言語処理学会第18回年次大会 優秀賞

    2012年9月2日 言語処理学会

  86. 日本経済新聞社「技術トレンド調査(2011年度第1回)」総合3位

    2011年5月30日 日本経済新聞社

  87. CICLing-2011 Best Paper Award (First Place)

    2011年2月20日 CICLing

  88. 言語処理学会第16回年次大会最優秀発表賞

    2010年8月 言語処理学会

  89. 言語処理学会第13回年次大会優秀発表賞

    2007年8月 言語処理学会

  90. マルチメディア・分散・協調とモバイルシンポジウム(DICOMO2006)野口賞

    2006年10月 情報処理学会他

  91. COLING/ACL-2006 Best Asian NLP Paper Award

    2006年7月 Association for Computational Linguistics

  92. 言語処理学会第10回年次大会優秀発表賞

    2004年8月 言語処理学会

  93. 言語処理学会第9回年次大会優秀発表賞

    2003年8月 言語処理学会

  94. 人工知能学会2000年度研究奨励賞

    2001年6月 人工知能学会

  95. 言語処理学会第4回年次大会優秀発表賞

    1998年8月 言語処理学会

  96. 人工知能学会1997年度論文賞

    1998年6月 人工知能学会

  97. 言語処理学会第2回年次大会優秀発表賞

    1996年8月 言語処理学会

︎全件表示 ︎最初の5件までを表示

論文 416

  1. Japanese-English Sentence Translation Exercises Dataset for Automatic Grading 査読有り

    Naoki Miura, Hiroaki Funayama, Seiya Kikuchi, Yuichiroh Matsubayashi, Yuya Iwase, Kentaro Inui

    266 (278) 2024年3月

  2. RealTime QA: What's the Answer Right Now? 査読有り

    Jungo Kasai, Keisuke Sakaguchi, Yoichi Takahashi, Ronan Le Bras, Akari Asai, Xinyan Velocity Yu, Dragomir Radev, Noah A. Smith, Yejin Choi, Kentaro Inui

    37th Conference on Neural Information Processing Systems (NeurIPS 2023) Track on Datasets and Benchmarks. 2023年12月

  3. Contrastive Learning-based Sentence Encoders Implicitly Weight Informative Words 査読有り

    Hiroto Kurita, Goro Kobayashi, Sho Yokoi, Kentaro Inui

    In Findings of the 2023 Conference on Empirical Methods in Natural Language Processing (EMNLP2023 Findings) 2023年12月

  4. Teach Me How to Argue: A Survey on NLP Feedback Systems in Argumentation 査読有り

    Camelia Guerraoui, Paul Reisert, Naoya Inoue, Farjana Sultana Mim, Keshav Singh, Jungmin Choi, Irfan Robbani, Shoichi Naito, Wenzhi Wang, Kentaro Inui

    Proceedings of the 10th Workshop on Argument Mining (ArgMining 2023) 19-34 2023年12月

    出版者・発行元:Association for Computational Linguistics

    DOI: 10.18653/v1/2023.argmining-1.3  

  5. Aoba_v3 bot: a multimodal chatbot system combining rules and various response generation models 査読有り

    Shoji Moriya, Daiki Shiono, Riki Fujihara, Yosuke Kishinami, Subaru Kimura, Shusaku Sone, Reina Akama, Yuta Matsumoto, Jun Suzuki, Kentaro Inui

    Advanced Robotics 1-14 2023年8月1日

    出版者・発行元:Informa UK Limited

    DOI: 10.1080/01691864.2023.2240883  

    ISSN:0169-1864

    eISSN:1568-5535

  6. Fight Bias with Bias? Two Interventions for Mitigating the Selective Avoidance of Clicking Uncongenial Facts 査読有り

    Yuko Tanaka, Hiromi Arai, Miwa Inuzuka, Yoichi Takahashi, Minao Kukita, Kentaro Inui

    Proceedings of the Annual Meeting of the Cognitive Science Society 45 2304-2310 2023年7月

  7. TohokuNLP at SemEval-2023 Task 5: Clickbait Spoiling via Simple Seq2seq Generation and Ensembling. 査読有り

    Hiroto Kurita, Ikumi Ito, Hiroaki Funayama, Shota Sasaki, Shoji Moriya, Ye Mengyu, Kazuma Kokuta, Ryujin Hatakeyama, Shusaku Sone, Kentaro Inui

    In Proceedings of the 17th International Workshop on Semantic Evaluation (SemEval) 1756-1762 2023年7月

  8. Transformer Language Models Handle Word Frequency in Prediction Head. 査読有り

    Goro Kobayashi, Tatsuki Kuribayashi, Sho Yokoi, Kentaro Inui

    In Findings of the 61st Annual Meeting of the Association for Computational Linguistics (ACL2023 Findings) 4523-4535 2023年7月

    DOI: 10.18653/v1/2023.findings-acl.276  

  9. Reducing the Cost: Cross-Prompt Pre-Finetuning for Short Answer Scoring. 査読有り

    Hiroaki Funayama, Yuya Asazuma, Yuichiroh Matsubayashi, Tomoya Mizumoto, Jun Suzuki, Kentaro Inui

    The 24th International Conference on Artificial Intelligence in Education (AIED2023) 78-89 2023年7月

    DOI: 10.1007/978-3-031-36272-9_7  

  10. Can LMs Store and Retrieve 1-to-N Relational Knowledge? 査読有り

    Haruki Nagasawa, Benjamin Heinzerling, Kazuma Kokuta, Kentaro Inui

    Archival submission for the ACL Student Research Workshop (ACL-SRW 2023), 130-138 2023年7月

  11. Examining the effect of whitening on static and contextualized word embeddings. 査読有り

    Shota Sasaki, Benjamin Heinzerling, Jun Suzuki, Kentaro Inui

    Information Processing & Management Vol.60 (No.3) 103272-103272 2023年5月

    DOI: 10.1016/j.ipm.2023.103272  

  12. Do Deep Neural Networks Capture Compositionality in Arithmetic Reasoning? 査読有り

    Keito Kudo, Yoichi Aoki, Tatsuki Kuribayashi, Ana Brassard, Masashi Yoshikawa, Keisuke Sakaguchi, Kentaro Inui

    In Proceedings of the 17th Conference of the European Chapter of the Association for Computational Linguistics (EACL2023) 1343-1354 2023年5月

    出版者・発行元:Association for Computational Linguistics

  13. Empirical Investigation of Neural Symbolic Reasoning Strategies. 査読有り

    Yoichi Aoki, Keito Kudo, Tatsuki Kuribayashi, Ana Brassard, Masashi Yoshikawa, Keisuke Sakaguchi, Kentaro Inui

    In Findings of the 17th Conference of the European Chapter of the Association for Computational Linguistics (EACL2023 Findings) 1124-1132 2023年5月

    出版者・発行元:Association for Computational Linguistics

  14. Prompting for explanations improves Adversarial NLI. Is this true? {Yes} it is {true} because {it weakens superficial cues}. 査読有り

    Pride Kavumba, Ana Brassard, Benjamin Heinzerling, Kentaro Inui

    In Proceedings of the 17th Conference of the European Chapter of the Association for Computational Linguistics (EACL2023) 2023年5月

  15. Who Does Not Benefit from Fact-checking Websites?

    Yuko Tanaka, Miwa Inuzuka, Hiromi Arai, Yoichi Takahashi, Minao Kukita, Kentaro Inui

    Proceedings of the 2023 CHI Conference on Human Factors in Computing Systems(CHI2023) 1-17 2023年4月19日

    出版者・発行元:ACM

    DOI: 10.1145/3544548.3580826  

  16. 自然言語処理×教育における説明能力 ―説明できるライティング評価技術への新しい展開―.

    乾健太郎, 石井雄隆, 松林優一郞, 井之上直也, 内藤昭一, 磯部順子, 舟山弘晃, 菊地正弥

    電子情報通信学会 基礎・境界ソサイエティ Fundamentals Review Vol.16 (No.4) 289-300 2023年4月

    ISSN:1882-0875

  17. Who Does Not Benefit from Fact-Checking Websites? A Psychological Characteristic Predicts the Selective Avoidance of Clicking Uncongenial Facts. 査読有り

    Yuko Tanaka, Miwa Inuzuka, Hiromi Arai, Yoichi Takahashi, Minao Kukita, Kentaro Inui

    In Proceedings of the 2023 CHI Conference on Human Factors in Computing Systems (CHI 2023) 2023年4月

  18. Use of an AI-powered Rewriting Support Software in Context with Other Tools: A Study of Non-Native English Speakers. 査読有り

    Takumi Ito, Naomi Yamashita, Tatsuki Kuribayashi, Masatoshi Hidaka, Jun Suzuki, Ge Gao, Jack Jamieson, Kentaro Inui

    UIST 45-13 2023年

    DOI: 10.1145/3586183.3606810  

  19. Investigating the Effectiveness of Multiple Expert Models Collaboration. 査読有り

    Ikumi Ito, Takumi Ito, Jun Suzuki, Kentaro Inui

    EMNLP (Findings) 14393-14404 2023年

  20. Test-time Augmentation for Factual Probing.

    Go Kamoda, Benjamin Heinzerling, Keisuke Sakaguchi, Kentaro Inui

    EMNLP (Findings) 3650-3661 2023年

  21. Feed-Forward Blocks Control Contextualization in Masked Language Models.

    Goro Kobayashi, Tatsuki Kuribayashi, Sho Yokoi, Kentaro Inui

    CoRR abs/2302.00456 2023年

    DOI: 10.48550/arXiv.2302.00456  

  22. Tracing and Manipulating Intermediate Values in Neural Math Problem Solvers. 査読有り

    Yuta Matsumoto, Benjamin Heinzerling, Masashi Yoshikawa, Kentaro Inui

    In The 1st Workshop on Mathematical Natural Language Processing (MathNLP 2022) 2022年12月

  23. Tracing and Manipulating Intermediate Results in Neural Math Problem Solvers. 査読有り

    Yuta Matsumoto, Benjamin Heinzerling, Masashi Yoshikawa, Kentaro Inui

    In Proceedings of the Fifth BlackboxNLP Workshop on Analyzing and Interpreting Neural Networks for NLP 2022年12月

  24. Cross-stitching Text and Knowledge Graph Encoders for Distantly Supervised Relation Extraction. 査読有り

    Qin Dai, Benjamin Heinzerling, Kentaro Inui

    In Proceedings of the 2022 Conference on Empirical Methods in Natural Language Processing (EMNLP 2022) 2022年12月

  25. Context Limitations Make Neural Language Models More Human-Like. 査読有り

    Tatsuki Kuribayashi, Yohei Oseki, Ana Brassard, Kentaro Inui

    In Proceedings of the 2022 Conference on Empirical Methods in Natural Language Processing (EMNLP 2022) 10421-10436 2022年12月

    出版者・発行元:Association for Computational Linguistics

    DOI: 10.18653/v1/2022.emnlp-main.712  

  26. 説明可能なAIを指向した和文英訳自動採点システムの開発と評価

    石井 雄隆, 菊地 正弥, 舟山 弘晃, 松林 優一郎, 乾 健太郎

    日本教育工学会研究報告集 2022 (4) 1-7 2022年11月28日

    出版者・発行元:一般社団法人 日本教育工学会

    DOI: 10.15077/jsetstudy.2022.4_1  

    eISSN:2436-3286

    詳細を見る 詳細を閉じる

    AIの判断結果の理由の説明や,品質を評価できる説明可能なAIについて近年盛んに議論されている.本研究では,説明可能なAIを指向した和文英訳自動採点システムの開発と評価を行った.このシステムでは,複数の評価観点を採点項目として反映したモデルにより自動採点を行い,学習者に診断的なフィードバックを行うことが可能となる.日本人大学生を対象とした刺激再生法を用いた実験の結果,システムを用いた修正の傾向やシステムの利点と改善点が明らかとなった.

  27. Chat Translation Error Detection for Assisting Cross-lingual Communications 査読有り

    Yunmeng Li, Jun Suzuki, Makoto Morishita, Kaori Abe, Ryoko Tokuhisa, Ana Brassard, Kentaro Inui

    the 3rd Workshop on Evaluation and Comparison of NLP Systems (Eval4NLP) 88-95 2022年11月

    DOI: 10.18653/v1/2022.eval4nlp-1.9  

  28. Empirical Investigation of Neural Symbolic Reasoning Strategies. 査読有り

    Yoichi Aoki, Keito Kudo, Tatsuki Kuribayashi, Ana Brassard, Masashi Yoshikawa, Keisuke Sakaguchi, Kentaro Inui

    Non-archival submission for the 2022 AACL-IJCNLP Student Research Workshop (AACL-IJCNLP SRW 2022) 1124-1132 2022年11月

    出版者・発行元:Association for Computational Linguistics

  29. Tracing and Manipulating Intermediate Results in Neural Math Problem Solvers. 査読有り

    Yuta Matsumoto, Benjamin Heinzerling, Masashi Yoshikawa, Kentaro Inui

    In Proceedings of the 2022 Conference on AACL-IJCNLP Student Research Workshop (AACL-IJCNLP SRW 2022) 2022年11月

  30. Why sentence similarity benchmark is not predictive of application-oriented task performance? 査読有り

    Kaori Abe, Yokoi Sho, Tomoyuki Kajiwara, Kentaro Inui

    In Proceedings of the 3rd Workshop on Evaluation & Comparison of NLP Systems (Eval4NLP 2022) 70-87 2022年11月

  31. 大学・大学院におけるレポート採点支援システムの開発. 査読有り

    佐々木健太, 鈴木健一, 君島朋子, 吉川将司, 乾健太郎

    情報処理学会論文誌教育とコンピュータ (TCE) Vol.8 (No.3) 35-49 2022年10月

  32. Target-Guided Open-Domain Conversation Planning. 査読有り

    Yosuke Kishinami, Reina Akama, Shiki Sato, Ryoko Tokuhisa, Jun Suzuki, Kentaro Inui

    In Proceedings of the 29th International Conference on Computational Linguistics (COLING 2022) 660-668 2022年10月

  33. Iterative Span Selection: Self-Emergence of Resolving Orders in Semantic Role Labeling. 査読有り

    Shuhei Kurita, Hiroki Ouchi, Kentaro Inui, Satoshi Sekine

    In Proceedings of the 29th International Conference onComputational Linguistics (COLING 2022) 5383-5397 2022年10月

  34. Topicalization in Language Models: A Case Study on Japanese. 査読有り

    Riki Fujihara, Tatsuki Kuribayashi, Kaori Abe, Kentaro Inui

    851-862 2022年10月

    出版者・発行元:None

  35. Analyzing Methods for Generating Feedback Comments for Language Learners. 査読有り

    Kazuaki Hanawa, Ryo Nagata, Kentaro Inui

    Journal of Natural Language Processing Volume 29 (No.3) 901-924 2022年9月

  36. N-best Response-based Analysis of Contradiction-awareness in Neural Response Generation Models. 査読有り

    Shiki Sato, Reina Akama, Hiroki Ouchi, Ryoko Tokuhisa, Jun Suzuki, Kentaro Inui

    In Proceedings of the 23rd Annual Meeting of the Special Interest Group on Discourse and Dialogue (SIGDIAL 2022) abs/2208.02578 637-644 2022年9月

    DOI: 10.48550/arXiv.2208.02578  

  37. Generating Feature Attribution-based Explanations for Automated Short Answer Scoring. 査読有り

    Tasuku Sato, Hiroaki Funayama, Kazuaki Hanawa, Kentaro Inui

    In Proceedings of the 23rd International Conference on Artificial Intelligence in Education (AIED2022) 231-242 2022年7月

  38. LPAttack: A Feasible Annotation Scheme for Capturing Logic Pattern of Attacks in Arguments. 査読有り

    Farjana Sultana Mim, Naoya Inoue, Shoichi Naito, Keshav Singh, Kentaro Inui

    In Proceedings of the 13th International Conference on the Language Resources and Evaluation Conference (LREC 2022) 2446-2459 2022年6月

  39. COPA-SSE: Semi-structured Explanations for Commonsense Reasoning. 査読有り

    Ana Brassard, Benjamin Heinzerling, Pride Kavumba, Kentaro Inui

    In Proceedings of the 13th International Conference on the Language Resources and Evaluation Conference (LREC 2022) 3994-4000 2022年6月

  40. IRAC: A Domain-specific Annotated Corpus of Implicit Reasoning in Arguments. 査読有り

    Keshav Singh, Naoya Inoue, Farjana Sultana Mim, Shoichi Naito, Kentaro Inui

    In Proceedings of the 13th International Conference on the Language Resources and Evaluation Conference (LREC 2022) 4674-4683 2022年6月

  41. TYPIC: A Corpus of Template-Based Diagnostic Comments on Argumentation. 査読有り

    Shoichi Naito, Shintaro Sawada, Chihiro Nakagawa, Naoya Inoue, Kenshi Yamaguchi, Iori Shimizu, Farjana Sultana Mim, Keshav Singh, Kentaro Inui

    In Proceedings of the 13th International Conference on the Language Resources and Evaluation Conference (LREC 2022) abs/2201.06674 5916-5928 2022年6月

  42. 負例を厳選した対話応答選択による対話応答生成システムの評価. 査読有り

    佐藤志貴, 赤間怜奈, 大内啓樹, 鈴木潤, 乾健太郎

    自然言語処理 Vol.29 (No.1) 53-83 2022年3月

    ISSN:2185-8314

  43. Toward mapping pragmatic impairment of autism spectrum disorder individuals through the development of a corpus of spoken Japanese Authors. 査読有り

    Sumi Kato, Kazuaki Hanawa, Vo Phuong Linh, Manabu Saito, Ryuichi Iimura, Kentaro Inui, Kazuhiko Nakamura

    Vol.17 (No.2) e0264204-e0264204 2022年2月

    出版者・発行元:None

    DOI: 10.1371/journal.pone.0264204  

    eISSN:1932-6203

  44. Context Limitations Make Neural Language Models More Human-Like.

    Tatsuki Kuribayashi, Yohei Oseki, Ana Brassard, Kentaro Inui

    CoRR abs/2205.11463 2022年

    DOI: 10.48550/arXiv.2205.11463  

  45. N-best Response-based Analysis of Contradiction-awareness in Neural Response Generation Models.

    Shiki Sato, Reina Akama, Hiroki Ouchi, Ryoko Tokuhisa, Jun Suzuki, Kentaro Inui

    SIGDIAL 637-644 2022年

  46. Evaluating Dialogue Response Generation Systems via Response Selection with Well-chosen False Candidates

    Shiki Sato, Reina Akama, Hiroki Ouchi, Jun Suzuki, Kentaro Inui

    Journal of Natural Language Processing 29 (1) 53-83 2022年

    出版者・発行元:Association for Natural Language Processing

    DOI: 10.5715/jnlp.29.53  

    ISSN:1340-7619

    eISSN:2185-8314

  47. Balancing Cost and Quality: An Exploration of Human-in-the-Loop Frameworks for Automated Short Answer Scoring. 査読有り

    Hiroaki Funayama, Tasuku Sato, Yuichiroh Matsubayashi, Tomoya Mizumoto, Jun Suzuki, Kentaro Inui

    AIED (1) 465-476 2022年

    出版者・発行元:Springer

    DOI: 10.1007/978-3-031-11644-5_38  

  48. Balancing Cost and Quality: An Exploration of Human-in-the-loop Frameworks for Automated Short Answer Scoring.

    Hiroaki Funayama, Tasuku Sato, Yuichiroh Matsubayashi, Tomoya Mizumoto, Jun Suzuki, Kentaro Inui

    CoRR abs/2206.08288 2022年

    DOI: 10.48550/arXiv.2206.08288  

  49. Diverse Lottery Tickets Boost Ensemble from a Single Pretrained Model.

    Sosuke Kobayashi, Shun Kiyono, Jun Suzuki, Kentaro Inui

    CoRR abs/2205.11833 2022年

    DOI: 10.48550/arXiv.2205.11833  

  50. Towards Automated Document Revision: Grammatical Error Correction, Fluency Edits, and Beyond.

    Masato Mita, Keisuke Sakaguchi, Masato Hagiwara, Tomoya Mizumoto, Jun Suzuki, Kentaro Inui

    CoRR abs/2205.11484 2022年

    DOI: 10.48550/arXiv.2205.11484  

  51. LPAttack: A Feasible Annotation Scheme for Capturing Logic Pattern of Attacks in Arguments.

    Farjana Sultana Mim, Naoya Inoue, Shoichi Naitoh, Keshav Singh, Kentaro Inui

    CoRR abs/2204.01512 2022年

    DOI: 10.48550/arXiv.2204.01512  

  52. Instance-Based Neural Dependency Parsing

    Hiroki Ouchi, Jun Suzuki, Sosuke Kobayashi, Sho Yokoi, Tatsuki Kuribayashi, Masashi Yoshikawa, Kentaro Inui

    Transactions of the Association for Computational Linguistics 9 1493-1507 2021年12月17日

    出版者・発行元:MIT Press - Journals

    DOI: 10.1162/tacl_a_00439  

    eISSN:2307-387X

    詳細を見る 詳細を閉じる

    Abstract Interpretable rationales for model predictions are crucial in practical applications. We develop neural models that possess an interpretable inference process for dependency parsing. Our models adopt instance-based inference, where dependency edges are extracted and labeled by comparing them to edges in a training set. The training edges are explicitly used for the predictions; thus, it is easy to grasp the contribution of each edge to the predictions. Our experiments show that our instance-based models achieve competitive accuracy with standard neural models and have the reasonable plausibility of instance-based explanations.

  53. 目標指向対話のためのサブゴール設計に基づく発話系列生成

    岸波 洋介, 赤間 怜奈, 佐藤 志貴, 徳久 良子, 鈴木 潤, 乾 健太郎

    人工知能学会研究会資料 言語・音声理解と対話処理研究会 93 70-73 2021年11月20日

    出版者・発行元:一般社団法人 人工知能学会

    DOI: 10.11517/jsaislud.93.0_70  

    ISSN:0918-5682

    eISSN:2436-4576

    詳細を見る 詳細を閉じる

    対話において説得や交渉,特定の話題への誘導を試みる際には,目標を達成できる自然な対話の流れを設計し,それに沿って能動的に対話を進めることが重要である.このとき,相手の発話は必ずしも事前に想定していた通りのものになるとは限らないため,一つの戦略として(1)ある時点から目標までの対話の流れを設計したうえで,(2)相手の発話に応じてその流れを更新していくことが考えられる.本研究では,誘導したい話題などの目標に向かって能動的に対話を進行可能な対話システム実現に向けて,まずは(1),つまりシステムが一人二役で目標を達成できる自然な対話の流れを設計する手法を提案する.具体的には,知識グラフを用いて目標を達成可能な道筋の候補を作成し,応答生成モデルにより,作成した道筋を辿りながら自然な発話系列を生成する.実験では,提案手法の有効性を「特定の話題へ誘導する対話の設計」という問題設定で検証する.

  54. Subword-Based Compact Reconstruction for Open-Vocabulary Neural Word Embeddings. 査読有り

    Shota Sasaki, Jun Suzuki, Kentaro Inui

    IEEE/ACM Transactions on Audio, Speech, and Language Processing Volume 29 3551-3564 2021年11月

    DOI: 10.1109/TASLP.2021.3125133  

  55. Corruption Is Not All Bad: Incorporating Discourse Structure Into Pre-Training via Corruption for Essay Scoring. 査読有り

    Farjana Sultana Mim, Naoya Inoue, Paul Reisert, Hiroki Ouchi, Kentaro Inui

    IEEE/ACM Transactions on Audio, Speech, and Language Processing Volume 29 2202-2215 2021年11月

    DOI: 10.1109/TASLP.2021.3088223  

  56. Exploring Methods for Generating Feedback Comments for Writing Learning. 査読有り

    Kazuaki Hanawa, Ryo Nagata, Kentaro Inui

    In Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing (EMNLP 2021) 9719-9730 2021年11月

  57. Shifted Absolute Position Embeddings for Transformers. 査読有り

    Shun Kiyono, Sosuke Kobayashi, Jun Suzuki, Kentaro Inui

    In Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing (EMNLP 2021) 3309-3321 2021年11月

  58. Incorporating Residual and Normalization Layers into Analysis of Masked Language Models. 査読有り

    Goro Kobayashi, Tatsuki Kuribayashi, Sho Yokoi, Kentaro Inui

    In Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing (EMNLP 2021) 4547-4568 2021年11月

    出版者・発行元:Association for Computational Linguistics

    DOI: 10.18653/v1/2021.emnlp-main.373  

  59. Summarize-then-Answer: Generating Concise Explanations for Multi-hop Reading Comprehension. 査読有り

    Naoya Inoue, Harsh Trivedi, Steven Sinha, Niranjan Balasubramanian, Kentaro Inui

    In Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing (EMNLP 2021) 6064-6080 2021年11月

    出版者・発行元:Association for Computational Linguistics

    DOI: 10.18653/v1/2021.emnlp-main.490  

  60. Transformer-based Lexically Constrained Headline Generation. 査読有り

    4085-4090 2021年11月

  61. Pseudo Zero Pronoun Resolution Improves Zero Anaphora Resolution. 査読有り

    Ryuto Konno, Shun Kiyono, Yuichiroh Matsubayashi, Hiroki Ouchi, Kentaro Inui

    In Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing (EMNLP 2021) abs/2104.07425 3790-3806 2021年11月

  62. Lower Perplexity is Not Always Human-Like. 査読有り

    2021年8月

  63. SyGNS: A Systematic Generalization Testbed Based on Natural Language Semantics. 査読有り

    2021年8月

  64. 訓練事例の影響の軽量な推定. 査読有り

    小林颯介, 横井祥, 鈴木潤, 乾健太郎

    自然言語処理 Volume 28 (No.2) 573-597 2021年6月

  65. 機械翻訳モデルの頑健性評価に向けた言語現象毎データセットの構築と分析. 査読有り

    藤井諒, 三田雅人, 阿部香央莉, 塙一晃, 森下睦, 鈴木潤, 乾健太郎

    自然言語処理 Volume 28 (Number 2) 450-478 2021年6月

    出版者・発行元:一般社団法人 言語処理学会

    DOI: 10.5715/jnlp.28.450  

    ISSN:1340-7619

    詳細を見る 詳細を閉じる

    <p>ニューラル機械翻訳 (NMT) の登場により,ニュース記事など文体の整った入力に対する翻訳の品質は著しく向上してきた.しかし,ソーシャル・ネットワーキング・サービス (SNS) に代表されるユーザ生成コンテンツ (UGC) を対象とした NMT の翻訳には依然として多くの課題が残されている.異文化・多言語交流の促進に向けた機械翻訳システムの活用には,そうした特異な入力を正確に扱うことのできる翻訳モデルの構築が不可欠である.近年では,UGC における翻訳品質の向上に向けたコンペティションが開催されるなどその重要性は広く認知されている.一方で,UGC に起因するどのような要因が機械翻訳システムの出力に悪影響を及ぼすのかは明らかでなく,偏在するユーザコンテンツの翻訳に向けた確かな方向性は依然として定まっていない.そこで本研究では,言語現象に着目した日英機械翻訳システムの頑健性測定データセット <b>PheMT</b> を提案する.特定の言語現象を含む文に特化したデータセットにより,当該表現の翻訳正解率,および正規化に基づく翻訳品質の差分を用いた精緻なエラー分析を可能にする.構築したデータセットを用いた評価により,広く商用に利用される機械翻訳システムを含む,最先端の NMT モデルにおいても十分に扱えない,対処すべき言語現象の存在を明らかにする.</p>

  66. Learning to Learn to be Right for the Right Reasons. 査読有り

    Pride Kavumba, Benjamin Heinzerling, Ana Brassard, Kentaro Inui

    In Proceedings of the 2021 Annual Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies (NAACL-HLT) 3890-3898 2021年6月

  67. Evaluation of Similarity-based Explanations. 査読有り

    Kazuaki Hanawa, Sho Yokoi, Satoshi Hara, Kentaro Inui

    In Proceedings of the Ninth International Conference on Learning Representations (ICLR 2021) 2021年5月

  68. Exploring Transitivity in Neural NLI Models through Veridicality. 査読有り

    Hitomi Yanaka, Koji Mineshima, Kentaro Inui

    In Proceedings of the 16th Conference of the European Chapter of the Association for Computational Linguistics (EACL) 920-934 2021年4月

  69. Language Models as Knowledge Bases: On Entity Representations, Storage Capacity, and Paraphrased Queries. 査読有り

    Benjamin Heinzerling, Kentaro Inui

    In Proceedings of the 16th Conference of the European Chapter of the Association for Computational Linguistics (EACL) 2021年4月

  70. Two Training Strategies for Improving Relation Extraction over Universal Graph. 査読有り

    Qin Dai, Naoya Inoue, Ryo Takahashi, Kentaro Inui

    In Proceedings of the 16th Conference of the European Chapter of the Association for Computational Linguistics (EACL) 2021年4月

  71. Data Augmentation by Rubrics for Short Answer Grading. 査読有り

    Tianqi Wang, Hiroki Ouchi, Hiroaki Funayama, Kentaro Inui

    The Association for Natural Language Processing Vol.28 (No.1) 2021年3月

  72. 文法誤り訂正モデルの横断評価. 査読有り

    三田雅人, 水本智也, 金子正弘, 永田亮, 乾健太郎

    自然言語処理 Volume 28 (Number 1) 2021年3月

  73. データ指向型アプローチによるクローズドブック質問応答. 査読有り

    鈴木正敏, 松田耕史, 大内啓樹, 鈴木潤, 乾健太郎

    自然言語処理 Volume 28 (Number 1) 3-25 2021年3月

    出版者・発行元:一般社団法人 言語処理学会

    DOI: 10.5715/jnlp.28.3  

    ISSN:1340-7619

    詳細を見る 詳細を閉じる

    <p> 近年の言語モデリング技術の進展に伴い,クローズドブック質問応答と呼ばれる,読解を伴わない質問応答タスクの研究が進められている.既存研究では,モデルの訓練可能なパラメータ数を増大させることや,モデルの構造をより洗練されることに重点が置かれてきた.本研究では,モデルに多様な事実知識を効率的に持たせるために,データ指向のアプローチを選択し,その有効性を検証する.具体的には,事実知識の知識源として Wikipedia を利用した拡張データセットを作成し,モデルの訓練に利用する.実験により,データ拡張を用いて訓練したモデルが,訓練データでは見られなかった質問に正しく答えることができ,モデルが拡張データから新しい知識を学習していることを示唆する結果が得られた.正解率の評価では,Quizbowl の従来の最高性能を更新し,TriviaQA の強力な系列生成モデルに匹敵する性能をおよそ 20 分の 1 のパラメータ数で実現した.</p>

  74. Incorporating Residual and Normalization Layers into Analysis of Masked Language Models.

    Goro Kobayashi, Tatsuki Kuribayashi, Sho Yokoi, Kentaro Inui

    CoRR abs/2109.07152 2021年

  75. Pseudo Zero Pronoun Resolution Improves Zero Anaphora Resolution.

    Ryuto Konno, Shun Kiyono, Yuichiroh Matsubayashi, Hiroki Ouchi, Kentaro Inui

    Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing 3790-3806 2021年

    出版者・発行元:Association for Computational Linguistics

    DOI: 10.18653/v1/2021.emnlp-main.308  

  76. 対話システムの先読み能力実現に向けた未来の展開まで生成する学習戦略の提案と分析

    岸波 洋介, 赤間 怜奈, 佐藤 志貴, 鈴木 潤, 徳久 良子, 乾 健太郎

    人工知能学会全国大会論文集 JSAI2021 3J2GS6b02-3J2GS6b02 2021年

    出版者・発行元:一般社団法人 人工知能学会

    DOI: 10.11517/pjsai.jsai2021.0_3j2gs6b02  

    詳細を見る 詳細を閉じる

    人間同士の対話において,ある時点での発話は,過去の文脈だけでなく未来の展開に動機付けられる場合がしばしばある.未来の展開を先読みし,それを現在の発話に活用することは,能動的な対話の進行に重要な要素のひとつであり,近年,ニューラル対話応答生成の研究領域においても未来の展開を先読みする能力は注目されつつある.我々はこれまでに対話システムの先読み能力を分析可能な「先読み雑談タスク」を提案した.本研究では,対話システムの先読み能力を実現するために,既存の対話システムの枠組みに対し未来の展開まで生成する学習戦略を提案する.具体的には,入出力がそれぞれ過去の文脈,未来の展開を含む複数発話で構成される対話データを用いて,応答とその先の未来の展開まで生成するsequence-to-sequenceモデルを構築する.実験では,提案戦略で学習した対話システムを用いて「先読み雑談タスク」を実施し,システムが先読みをおこなえるのか分析する.実験結果から,既存の対話システムの枠組みで先読みをおこなう難しさを考察する.

  77. Annotating Implicit Reasoning in Arguments with Causal Links.

    Keshav Singh, Naoya Inoue, Farjana Sultana Mim, Shoichi Naitoh, Kentaro Inui

    CoRR abs/2110.13692 2021年

  78. A Comparative Study on Collecting High-Quality Implicit Reasonings at a Large-scale.

    Keshav Singh, Paul Reisert, Naoya Inoue, Kentaro Inui

    CoRR abs/2104.07924 2021年

  79. Exploring Methodologies for Collecting High-Quality Implicit Reasoning in Arguments.

    Keshav Singh, Farjana Sultana Mim, Naoya Inoue, Shoichi Naitoh, Kentaro Inui

    Proceedings of the 8th Workshop on Argument Mining(ArgMining@EMNLP) 57-66 2021年

    出版者・発行元:Association for Computational Linguistics

    DOI: 10.18653/v1/2021.argmining-1.6  

  80. Towards AI Systems That Can Explain with Language

    Kentaro Inui, Daisuke Bekki, Sadao Kurohashi, Minao Kukita

    Journal of Natural Language Processing 28 (3) 874-880 2021年

    出版者・発行元:Association for Natural Language Processing

    DOI: 10.5715/jnlp.28.874  

    ISSN:1340-7619

    eISSN:2185-8314

  81. Instance-Based Neural Dependency Parsing.

    Hiroki Ouchi, Jun Suzuki, Sosuke Kobayashi, Sho Yokoi, Tatsuki Kuribayashi, Masashi Yoshikawa, Kentaro Inui

    CoRR abs/2109.13497 2021年

  82. Cross-Sectional Evaluation of Grammatical Error Correction Models

    Masato Mita, Tomoya Mizumoto, Masahiro Kaneko, Ryo Nagata, Kentaro Inui

    Journal of Natural Language Processing 28 (1) 160-182 2021年

    出版者・発行元:Association for Natural Language Processing

    DOI: 10.5715/jnlp.28.160  

    ISSN:1340-7619

    eISSN:2185-8314

  83. 論述構造解析におけるスパン分散表現. 査読有り

    栗林樹生, 大内啓樹, 井之上直也, 鈴木潤, Paul Reisert, 三好利昇, 乾健太郎

    自然言語処理 Volume 27 (Number 4) 753-780 2020年12月

  84. Multi-dialect Neural Machine Translation for 48 Low-resource Japanese Dialects. 査読有り

    Kaori Abe, Yuichiroh Matsubayashi, Naoaki Okazaki, Kentaro Inui

    自然言語処理 Volume 27 (Number 4) 781-800 2020年12月

    出版者・発行元:None

    DOI: 10.5715/jnlp.27.781  

    ISSN:1340-7619

    eISSN:2185-8314

  85. PheMT: A Phenomenon-wise Dataset for Machine Translation Robustness on User-Generated Contents. 査読有り

    Ryo Fujii, Masato Mita, Kaori Abe, Kazuaki Hanawa, Makoto Morishita, Jun Suzuki, Kentaro Inui

    In Proceedings of the 28th International Conference on Computational Linguistics (COLING 2020) 5929-5943 2020年12月

    DOI: 10.18653/v1/2020.coling-main.521  

  86. An Empirical Study of Contextual Data Augmentation for Japanese Zero Anaphora Resolution. 査読有り

    Ryuto Konno, Yuichiroh Matsubayashi, Shun Kiyono, Hiroki Ouchi, Ryo Takahashi, Kentaro Inui

    In Proceedings of the 28th International Conference on Computational Linguistics (COLING 2020) 4956-4968 2020年12月

    出版者・発行元:None

    DOI: 10.18653/v1/2020.coling-main.435  

  87. Modeling Event Salience in Narratives via Barthes’ Cardinal Functions. 査読有り

    Takaki Otake, Sho Yokoi, Naoya Inoue, Ryo Takahashi, Tatsuki Kuribayashi, Kentaro Inui

    In Proceedings of the 28th International Conference on Computational Linguistics (COLING 2020) 1784-1794 2020年12月

    出版者・発行元:None

    DOI: 10.18653/v1/2020.coling-main.160  

  88. Efficient Estimation of Influence of a Training Instance. 査読有り

    Sosuke Kobayashi, Sho Yokoi, Jun Suzuki, Kentaro Inui

    In Proceedings of EMNLP 2020 Workshop Workshop on Simple and Efficient Natural Language Processing (SustaiNLP) 41-47 2020年11月

    出版者・発行元:None

    DOI: 10.18653/v1/2020.sustainlp-1.6  

  89. Word Rotator’s Distance. 査読有り

    Sho Yokoi, Ryo Takahashi, Reina Akama, Jun Suzuki, Kentaro Inui

    In Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing (EMNLP 2020) 2944-2960 2020年11月

    DOI: 10.18653/v1/2020.emnlp-main.236  

  90. Filtering Noisy Dialogue Corpora by Connectivity and Content Relatedness. 査読有り

    Reina Akama, Sho Yokoi, Jun Suzuki, Kentaro Inui

    In Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing (EMNLP 2020) 941-958 2020年11月

    DOI: 10.18653/v1/2020.emnlp-main.68  

  91. Langsmith: An Interactive Academic Text Revision System. 査読有り

    Takumi Ito, Tatsuki Kuribayashi, Masatoshi Hidaka, Jun Suzuki, Kentaro Inui

    In Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing: System Demonstrations (EMNLP 2020) 216-226 2020年11月

    出版者・発行元:None

    DOI: 10.18653/v1/2020.emnlp-demos.28  

  92. Attention is Not Only a Weight: Analyzing Transformers with Vector Norms. 査読有り

    Goro Kobayashi, Tatsuki Kuribayashi, Sho Yokoi, Kentaro Inui

    In Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing (EMNLP 2020) 7057-7075 2020年11月

    出版者・発行元:None

    DOI: 10.18653/v1/2020.emnlp-main.574  

  93. A Self-Refinement Strategy for Noise Reduction in Grammatical Error Correction. 査読有り

    Masato Mita, Shun Kiyono, Masahiro Kaneko, Jun Suzuki, Kentaro Inui

    In Findings of the 2020 Conference on Empirical Methods in Natural Language Processing (EMNLP 2020) 267-280 2020年11月

    DOI: 10.18653/v1/2020.findings-emnlp.26  

  94. Preventing Critical Scoring Errors in Short Answer Scoring with Confidence Estimation. 査読有り

    Hiroaki Funayama, Shota Sasaki, Yuichiroh Matsubayashi, Tomoya Mizumoto, Jun Suzuki, Masato Mita, Kentaro Inui

    In Proceedings of the 2020 ACL Student Research Workshop (2020 ACL SRW) 237-243 2020年7月

    出版者・発行元:None

    DOI: 10.18653/v1/2020.acl-srw.32  

  95. Embeddings of Label Components for Sequence Labeling: A Case Study of Fine-grained Named Entity Recognition. 査読有り

    Takuma Kato, Kaori Abe, Hiroki Ouchi, Shumpei Miyawaki, Jun Suzuki, Kentaro Inui

    In Proceedings of the 2020 ACL Student Research Workshop (2020 ACL SRW) 222-229 2020年7月

    DOI: 10.18653/v1/2020.acl-srw.30  

  96. Self-Attention is Not Only a Weight: Analyzing BERT with Vector Norms. 査読有り

    Goro Kobayashi, Tatsuki Kuribayashi, Sho Yokoi, Kentaro Inui

    Non-archival submission for the 2020 ACL Student Research Workshop (2020 ACL SRW) 7057-7075 2020年7月

    出版者・発行元:None

    DOI: 10.18653/v1/2020.emnlp-main.574  

  97. R4C: A Benchmark for Evaluating RC Systems to Get the Right Answer for the Right Reason. 査読有り

    Naoya Inoue, Pontus Stenetorp, Kentaro Inui

    In Proceedings of the 58th annual meeting of the Association for Computational Linguistics (ACL 2020) 6740-6750 2020年7月

    出版者・発行元:None

    DOI: 10.18653/v1/2020.acl-main.602  

  98. Instance-based Learning of Span Representations: A Case Study through Named Entity Recognition. 査読有り

    Hiroki Ouchi, Jun Suzuki, Sosuke Kobayashi, Sho Yokoi, Tatsuki Kuribayashi, Ryuto Konno, Kentaro Inui

    In Proceedings of the 58th annual meeting of the Association for Computational Linguistics (ACL 2020) abs/2004.14514 6452-6459 2020年7月

  99. Language Models as an Alternative Evaluator of Word Order Hypotheses: A Case Study in Japanese. 査読有り

    Tatsuki Kuribayashi, Takumi Ito, Jun Suzuki, Kentaro Inui

    In Proceedings of the 58th annual meeting of the Association for Computational Linguistics (ACL 2020) abs/2005.00842 488-504 2020年7月

  100. Evaluating Dialogue Generation Systems via Response Selection. 査読有り

    Shiki Sato, Reina Akama, Hiroki Ouchi, Jun Suzuki, Kentaro Inui

    In Proceedings of the 58th annual meeting of the Association for Computational Linguistics (ACL 2020) abs/2004.14302 593-599 2020年7月

  101. Can Encoder-decoder Models Benefit from Pre-trained LanguageRepresentation in Grammatical Error Correction?. 査読有り

    Masahiro Kaneko, Masato Mita, Shun Kiyono, Jun Suzuki, Kentaro Inui

    In Proceedings of the 58th annual meeting of the Association for Computational Linguistics (ACL 2020) 4248-4254 2020年7月

    出版者・発行元:None

    DOI: 10.18653/v1/2020.acl-main.391  

  102. Do Neural Models Learn Systematicity of Monotonicity Inference in Natural Language?. 査読有り

    Hitomi Yanaka, Koji Mineshima, Daisuke Bekki, Kentaro Inui

    In Proceedings of the 58th annual meeting of the Association for Computational Linguistics (ACL 2020) 6105-6117 2020年7月

  103. An empirical exploration of the challenges in temporal relation extraction from clinical text. 査読有り

    Diana Galvan-Sosa, Koji Matsuda, Naoaki Okazaki, Kentaro Inui

    Journal of Natural Language Processing, Vol.27 (No.2) 2020年6月

  104. Creating Corpora for Research in Feedback Comment Generation. 査読有り

    Ryo Nagata, Kentaro Inui, Shin’ichiro Ishikawa

    In Proceedings of the 12th International Conference on the Language Resources and Evaluation Conference (LREC 2020) 340-345 2020年5月

    出版者・発行元:None

  105. Assessing the Benchmarking Capacity of Machine Reading Comprehension Datasets. 査読有り

    Saku Sugawara, Pontus Stenetorp, Kentaro Inui, Akiko Aizawa

    The 34th AAAI Conference on Artificial Intelligence (AAAI-20) 8918-8927 2020年2月

    出版者・発行元:None

  106. Do Neural Models Learn Systematicity of Monotonicity Inference in Natural Language?

    Hitomi Yanaka, Koji Mineshima, Daisuke Bekki, Kentaro Inui

    CoRR abs/2004.14839 2020年

  107. Modeling Event Salience in Narratives via Barthes' Cardinal Functions.

    Takaki Otake, Sho Yokoi, Naoya Inoue, Ryo Takahashi, Tatsuki Kuribayashi, Kentaro Inui

    CoRR abs/2011.01785 2020年

  108. Language Models as an Alternative Evaluator of Word Order Hypotheses: A Case Study in Japanese.

    Tatsuki Kuribayashi, Takumi Ito, Jun Suzuki, Kentaro Inui

    Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics(ACL) 488-504 2020年

    出版者・発行元:Association for Computational Linguistics

    DOI: 10.18653/v1/2020.acl-main.47  

  109. A System for Worldwide COVID-19 Information Aggregation.

    Akiko Aizawa, Frédéric Bergéron, Junjie Chen, Fei Cheng, Katsuhiko Hayashi, Kentaro Inui, Hiroyoshi Ito, Daisuke Kawahara, Masaru Kitsuregawa, Hirokazu Kiyomaru, Masaki Kobayashi, Takashi Kodama, Sadao Kurohashi, Qianying Liu, Masaki Matsubara, Yusuke Miyao, Atsuyuki Morishima, Yugo Murawaki, Kazumasa Omura, Haiyue Song, Eiichiro Sumita, Shinji Suzuki, Ribeka Tanaka, Yu Tanaka, Masashi Toyoda, Nobuhiro Ueda, Honai Ueoka, Masao Utiyama, Ying Zhong

    2020年

    DOI: 10.18653/v1/2020.nlpcovid19-2.13  

  110. Instance-Based Learning of Span Representations: A Case Study through Named Entity Recognition.

    Hiroki Ouchi, Jun Suzuki, Sosuke Kobayashi, Sho Yokoi, Tatsuki Kuribayashi, Ryuto Konno, Kentaro Inui

    Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics(ACL) 6452-6459 2020年

    出版者・発行元:Association for Computational Linguistics

    DOI: 10.18653/v1/2020.acl-main.575  

  111. Encoder-Decoder Models Can Benefit from Pre-trained Masked Language Models in Grammatical Error Correction.

    Masahiro Kaneko, Masato Mita, Shun Kiyono, Jun Suzuki, Kentaro Inui

    CoRR abs/2005.00987 2020年

  112. Word Rotator's Distance: Decomposing Vectors Gives Better Representations.

    Sho Yokoi, Ryo Takahashi, Reina Akama, Jun Suzuki, Kentaro Inui

    CoRR abs/2004.15003 2020年

  113. Utterance Pair Scoring for Noisy Dialogue Data Filtering.

    Reina Akama, Sho Yokoi, Jun Suzuki, Kentaro Inui

    CoRR abs/2004.14008 2020年

  114. A System for Worldwide COVID-19 Information Aggregation.

    Akiko Aizawa, Frédéric Bergéron, Junjie Chen, Fei Cheng, Katsuhiko Hayashi, Kentaro Inui, Hiroyoshi Ito, Daisuke Kawahara, Masaru Kitsuregawa, Hirokazu Kiyomaru, Masaki Kobayashi, Takashi Kodama, Sadao Kurohashi, Qianying Liu, Masaki Matsubara, Yusuke Miyao, Atsuyuki Morishima, Yugo Murawaki, Kazumasa Omura, Haiyue Song, Eiichiro Sumita, Shinji Suzuki, Ribeka Tanaka, Yu Tanaka, Masashi Toyoda, Nobuhiro Ueda, Honai Ueoka, Masao Utiyama, Ying Zhong

    CoRR abs/2008.01523 2020年

  115. Massive Exploration of Pseudo Data for Grammatical Error Correction. 査読有り

    Shun Kiyono, Jun Suzuki, Tomoya Mizumoto, Kentaro Inui

    IEEE/ACM Transactions on Audio, Speech, and Language Processing Vol.28 2134-2145 2020年

    DOI: 10.1109/TASLP.2020.3007753  

  116. Assisting Authors to Convert Raw Products into Polished Prose. 査読有り

    Takumi Ito, Tatsuki Kuribayashi, Hayato Kobayashi, Ana Brassard, Masato Hagiwara, Jun Suzuki, Kentaro Inui

    Journal of Cognitive Science Vol.21 (No.1) 101-137 2020年

  117. Improving Evidence Detection by Leveraging Warrants. 査読有り

    Keshav Singh, Paul Reisert, Naoya Inoue, Pride Kavumba, Kentaro Inui

    Proceedings of FEVER: The Second Workshop on Fact Extraction and Verification 57-62 2019年11月

  118. Inject Rubrics Into Short Answer Grading System. 査読有り

    Tianqi Wang, Naoya Inoue, Hiroki Ouchi, Tomoya Mizumoto, Kentaro Inui

    Proceedings of the 2nd Workshop on Deep Learning Approaches for Low-Resource NLP (DeepLo 2019) 175-182 2019年11月

    出版者・発行元:None

    DOI: 10.18653/v1/D19-6119  

  119. When Choosing Plausible Alternatives, Clever Hans can be Clever. 査読有り

    Pride Kavumba, Naoya Inoue, Benjamin Heinzerling, Keshav Singh, Paul Reisert, Kentaro Inui

    Proceedings of the First Workshop on Commonsense Inference in Natural Language Processing 33-42 2019年11月

  120. An Empirical Study of Incorporating Pseudo Data into Grammatical Error Correction. 査読有り

    Shun Kiyono, Jun Suzuki, Masato Mita, Tomoya Mizumoto, Kentaro Inui

    In Proceedings of 2019 Conference on Empirical Methods in Natural Language Processing and 9th International Joint Conference on Natural Language Processing (EMNLP-IJCNLP 2019) 1236-1242 2019年11月

    DOI: 10.18653/v1/D19-1119  

  121. Transductive Learning of Neural Language Models for Syntactic and Semantic Analysis. 査読有り

    Hiroki Ouchi, Jun Suzuki, Kentaro Inui

    In Proceedings of 2019 Conference on Empirical Methods in Natural Language Processing and 9th International Joint Conference on Natural Language Processing (EMNLP-IJCNLP 2019) 3665-3671 2019年11月

    出版者・発行元:None

    DOI: 10.18653/v1/D19-1379  

  122. Select and Attend: Towards Controllable Content Selection in Text Generation. 査読有り

    Xiaoyu Shen, Jun Suzuki, Kentaro Inui, Hui Su, Dietrich Klakow, Satoshi Sekine

    In Proceedings of 2019 Conference on Empirical Methods in Natural Language Processing and 9th International Joint Conference on Natural Language Processing (EMNLP-IJCNLP 2019) 579-590 2019年11月

    DOI: 10.18653/v1/D19-1054  

  123. TEASPN: Framework and Protocol for Integrated Writing Assistance Environments. 査読有り

    Masato Hagiwara, Takumi Ito, Tatsuki Kuribayashi, Jun Suzuki, Kentaro Inui

    To appear in Proceedings of 2019 Conference on Empirical Methods in Natural Language Processing and 9th International Joint Conference on Natural Language Processing: System Demonstrations (EMNLP-IJCNLP 2019) 229-234 2019年11月

    出版者・発行元:None

    DOI: 10.18653/v1/D19-3039  

  124. A Large-Scale Multi-Length Headline Corpus for Analyzing Length-Constrained Headline Generation Model Evaluation. 査読有り

    Yuta Hitomi, Yuya Taguchi, Hideaki Tamori, Ko Kikuta, Jiro Nishitoba, Naoaki Okazaki, Kentaro Inui, Manabu Okumura

    In Proceedings of the 12th International Conference on Natural Language Generation (INLG 2019) 333-343 2019年11月

  125. Diamonds in the Rough: Generating Fluent Sentences from Early-stage Drafts for Academic Writing Assistance. 査読有り

    Takumi Ito, Tatsuki Kuribayashi, Hayato Kobayashi, Ana Brassard, Masato Hagiwara, Jun Suzuki, Kentaro Inui

    To appear in Proceedings of the 12th International Conference on Natural Language Generation (INLG 2019) 40-53 2019年10月

    出版者・発行元:None

    DOI: 10.18653/v1/W19-8606  

  126. Incorporating Chains of Reasoning over Knowledge Graph for Distantly Supervised Biomedical Knowledge Acquisition. 査読有り

    Qin Dai, Naoya Inoue, Paul Reisert, Ryo Takahashi, Kentaro Inui

    In Proceedings of the 33rd Pacific Asia Conference on Language, Information and Computation (PACLIC 33) 19-28 2019年9月

  127. Quality of Life情報の伝達補助を目的とする対話応答候補生成. 査読有り

    赤間 怜奈, 徳久 良子, 乾健太郎

    自然言語処理 Volume 26 (Number 3) 579-612 2019年9月

  128. クイズ解答タスクにおける大規模ラベルなしコーパスの利用: 言語モデルとデータ拡張.

    鈴木正敏, 松田耕史, 大内啓樹, 鈴木潤, 乾健太郎

    第14回NLP若手の会 シンポジウム (YANS) 2019 (NL-241) 2019年8月

  129. 文ベクトルの最適輸送に基づく類似性尺度.

    横井祥, 高橋諒, 赤間怜奈, 鈴木潤, 乾健太郎

    第14回NLP若手の会 シンポジウム (YANS) 2019年8月

  130. Analytic Score Prediction and Justification Identification in Automated Short Answer Scoring. 査読有り

    Tomoya Mizumoto, Hiroki Ouchi, Yoriko Isobe, Paul Reisert, Ryo Nagata, Satoshi Sekine, Kentaro Inui

    316-325 2019年8月

    DOI: 10.18653/v1/w19-4433  

  131. 活動目的に基づく危機対応ログ分類による危機対応マネジメント支援手法の検討.

    小阪 尚子, 爰川 知宏, 山口 健史, 乾 健太郎

    情報処理学会 マルチメディア,分散,協調とモバイル シンポジウム 2019 (1) 1772-1777 2019年7月

    ISSN:1882-0840

  132. An Empirical Study of Span Representations in Argumentation Structure Parsing. 査読有り

    Tatsuki Kuribayashi, Hiroki Ouchi, Naoya Inoue, Paul Reisert, Toshinori Miyoshi, Jun Suzuki, Kentaro Inui

    In Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics (ACL 2019) 4691-4698 2019年7月

    出版者・発行元:None

    DOI: 10.18653/v1/p19-1464  

  133. Unsupervised Learning of Discourse-Aware Text Representation for Essay Scoring. 査読有り

    Farjana Sultana Mim, Naoya Inoue, Paul Reisert, Hiroki Ouchi, Kentaro Inui

    In Proceedings of the 57th Conference of the Association for Computational Linguistics: Student Research Workshop (ACL SRW 2019) 378-385 2019年7月

    出版者・発行元:None

    DOI: 10.18653/v1/p19-2053  

  134. Can neural networks understand monotonicity reasoning? 査読有り

    Hitomi Yanaka, Koji Mineshima, Daisuke Bekki, Kentaro Inui, Satoshi Sekine, Lasha Abzianidze, Johan Bos

    In Proceedings of ACL2019 Workshop on Analyzing and interpreting neural networks for NLP (BlackboxNLP2019) abs/1906.06448 31-40 2019年7月

  135. An Annotation Protocol for Collecting User-Generated Counter-Arguments using Crowdsourcing. 査読有り

    Paul Reisert, Gisela Vallejo, Naoya Inoue, Iryna Gurevych, Kentaro Inui

    in Proceedings of the 20th International Conference on Artificial Intelligence in Education (AIED) 2019年7月

  136. 機械翻訳における訳語一貫性評価用データセットの構築.

    阿部香央莉, 鈴木潤, 永田昌明, 乾健太郎

    2019年度人工知能学会全国大会(第33回) 33rd 2019年6月

  137. Overview of the NTCIR-14 QA Lab-PoliInfo Task. 査読有り

    Yasutomo Kimura, Hideyuki Shibuki, Hokuto Ototake, Yuzu Uchida, Keiichi Takamaru, Kotaro Sakamoto, Madoka Ishioroshi, Teruko Mitamura, Noriko Kando, Tatsunori Mori, Harumichi Yuasa, Satoshi Sekine, Kentaro Inui

    In Proceedings of the 14th NTCIR Conference 2019年6月

  138. Distantly Supervised Biomedical Knowledge Acquisition via Knowledge Graph Based Attention. 査読有り

    Qin Dai, Naoya Inoue, Paul Reisert, Ryo Takahashi, Kentaro Inui

    In Proceedings of the Workshop on Extracting Structured Knowledge from Scientific Publications 1-10 2019年6月

  139. 単語埋め込みにおける複数視点の対義語判定.

    中村拓, 田然, 佐々木翔大, 乾健太郎

    2019年度人工知能学会全国大会(第33回) 2019年6月

  140. HELP: A Dataset for Identifying Shortcomings of Neural Models in Monotonicity Reasoning. 査読有り

    Hitomi Yanaka, Koji Mineshima, Daisuke Bekki, Kentaro Inui, Satoshi Sekine, Lasha Abzianidze, Johan Bos

    In Proceedings of the 8th Joint Conference on Lexical and Computational Semantics (*SEM), 250-255 2019年6月

    出版者・発行元:None

    DOI: 10.18653/v1/s19-1027  

  141. Subword-based Compact Reconstruction of Word Embeddings. 査読有り

    Shota Sasaki, Jun Suzuki, Kentaro Inui

    Proceedings of the 17th Annual Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies (NAACL-HLT 2019) 3498-3508 2019年6月

    DOI: 10.18653/v1/n19-1353  

  142. The Sally Smedley Hyperpartisan News Detector at SemEval-2019 Task 4. 査読有り

    Kazuaki Hanawa, Shota Sasaki, Hiroki Ouchi, Jun Suzuki, Kentaro Inui

    Proceedings of the 13th International Workshop on Semantic Evaluation (SemEval) 1057-1061 2019年6月

    DOI: 10.18653/v1/s19-2185  

  143. ニューラル自然言語推論に向けたMonotonicityに基づく含意関係認識データセット自動構築.

    谷中瞳, 峯島宏次, 戸次大介, 関根聡, 乾健太郎, Lasha Abzianidze, Johan Bos

    言語処理学会第25回年次大会 1527-1530 2019年3月

  144. Unsupervised Learning of Discourse-Aware Text Representation.

    Farjana Sultana Mim, Naoya Inoue, Paul Reisert, Hiroki Ouchi, Kentaro Inui

    言語処理学会第25回年次大会 1471-1474 2019年3月

  145. ニューラル機械翻訳における文脈情報の選択的利用.

    藤井諒, 清野舜, 鈴木潤, 乾健太郎

    言語処理学会第25回年次大会 25th 1459-1462 2019年3月

    ISSN:2188-4420

  146. クラウドソーシングによるパーラメンタリーディベートへの論述構造のアノテーション.

    内藤昭一, 井之上直也, 乾健太郎

    言語処理学会第25回年次大会 1304-1307 2019年3月

  147. 出力長制御を考慮した見出し生成モデルのための大規模コーパス.

    人見雄太, 田口雄哉, 田森秀明, 菊田洸, 西鳥羽二郎, 岡崎直観, 乾健太郎, 奥村学

    言語処理学会第25回年次大会 1125-1128 2019年3月

  148. 計算機科学論文における手法の利点・欠点に着目したデータの構築と分析.

    白井穂乃, 井之上直也, 鈴木潤, 乾健太郎

    言語処理学会第25回年次大会 25th 1181-1184 2019年3月

    ISSN:2188-4420

  149. 独立発話の繋ぎ合わせによる発話-応答ペアの獲得.

    赤間怜奈, 武藤由依, 鈴木潤, 乾健太郎

    言語処理学会第25回年次大会 1153-1156 2019年3月

  150. 言語モデルを用いた日本語の語順評価と基本語順の分析.

    栗林樹生, 伊藤拓海, 内山香, 鈴木潤, 乾健太郎

    言語処理学会第25回年次大会 25th 1053-1056 2019年3月

    ISSN:2188-4420

  151. 強化学習によるプログラム?成のためのプログラム系列分析.

    佐藤拓海, 大内啓樹, 松田耕史, 鈴木正敏, 鈴木潤, 乾健太郎

    言語処理学会第25回年次大会 25th 1010-1013 2019年3月

    ISSN:2188-4420

  152. ExpertとImitatorの混合ネットワークによる大規模半教師あり学習.

    清野舜, 鈴木潤, 乾健太郎

    言語処理学会第25回年次大会 25th 1006-1009 2019年3月

    ISSN:2188-4420

  153. 複数の言語単位に対するスパン表現を用いた論述構造解析.

    栗林樹生, 大内啓樹, 井之上直也, Paul Reisert, 三好利昇, 鈴木潤, 乾健太郎

    言語処理学会第25回年次大会 25th 990-993 2019年3月

    ISSN:2188-4420

  154. 文法誤り訂正のコーパス横断評価:単一コーパス評価で十分か?.

    三田雅人, 水本智也, 金子正弘, 永田亮, 乾健太郎

    言語処理学会第25回年次大会 978-981 2019年3月

  155. ライティング支援を想定した情報補完型生成.

    伊藤拓海, 栗林樹生, 小林隼人, 鈴木潤, 乾健太郎

    言語処理学会第25回年次大会 25th 970-973 2019年3月

    ISSN:2188-4420

  156. フレーズ単位の発話応答ペアを用いた対話応答生成の多様化.

    佐藤志貴, 大内啓樹, 井之上直也, 鈴木潤, 乾健太郎

    言語処理学会第25回年次大会 25th 966-969 2019年3月

    ISSN:2188-4420

  157. 文法誤り訂正における反復訂正の効果検証.

    浅野広樹, 鈴木潤, 水本智也, 乾健太郎

    言語処理学会第25回年次大会 25th 578-581 2019年3月

    ISSN:2188-4420

  158. 採点基準を利用した記述式答案の自動採点.

    王天奇, 井之上直也, 水本智也, 大内啓樹, 乾健太郎

    言語処理学会第25回年次大会 450-453 2019年3月

  159. 顕現的要素の出現順序に基づく物語の類似性尺度.

    大竹孝樹, 横井祥, 井之上直也, 乾健太郎

    言語処理学会第25回年次大会 304-307 2019年3月

  160. 確率木置換文法と分散表現に基づくレシピ構造木の生成モデルの学習.

    吉成未菜里, 横井祥, 進藤裕之, 乾健太郎

    言語処理学会第25回年次大会 300-303 2019年3月

  161. サブワードに基づく単語分散表現の縮約モデリング.

    佐々木翔大, 鈴木潤, 乾健太郎

    言語処理学会第25回年次大会 25th 213-216 2019年3月

    ISSN:2188-4420

  162. 意味役割付与におけるトランズダクティブ分野適応.

    大内啓樹, 鈴木潤, 乾健太郎

    言語処理学会第25回年次大会 25th 155-158 2019年3月

    ISSN:2188-4420

  163. A Crowdsourceable Protocol for Collecting User-Generated Counter-Arguments.

    Paul Reisert, Naoya Inoue, Kentaro Inui

    言語処理学会第25回年次大会, 151-154 2019年3月

  164. End-to-End Scientific Knowledge Graph Completion via Word Embedding based Entity Type Classification.

    Qin Dai, Naoya Inoue, Paul Reisert, Kentaro Inui

    言語処理学会第25回年次大会 77-80 2019年3月

  165. 画像/言語同時埋め込みベクトル空間の構築に向けた埋め込み粒度の比較検討.

    北山晃太郎, 清野舜, 鈴木潤, 乾健太郎

    言語処理学会第25回年次大会 25th 1419-1422 2019年3月

    ISSN:2188-4420

  166. Improving Evidence Detection using Warrants as External Knowledge.

    Keshav Singh, Naoya Inoue, Paul Reisert, Kentaro Inui

    言語処理学会第25回年次大会 1241-1244 2019年3月

  167. A Crowdsourceable Protocol for Annotating Multi-Hop QA with Reasoning Steps.

    Naoya Inoue, Pontus Stenetorp, Kentaro Inui

    言語処理学会第25回年次大会 994-997 2019年3月

  168. Wikidata からの遠距離教師あり学習に基づく大規模関係知識獲得.

    松田耕史, 鈴木正敏, 乾健太郎

    言語処理学会第25回年次大会 659-662 2019年3月

  169. Exploring Supervised Learning of Hierarchical Event Embedding with Poincaré Embeddings.

    Pride Kavumba, Naoya Inoue, Kentaro Inui

    言語処理学会第25回年次大会 217-220 2019年3月

  170. 前方文脈の埋め込みを利用した日本語述語項構造解析.

    今野颯人, 松林優一郎, 大内啓樹, 清野舜, 乾健太郎

    言語処理学会第25回年次大会 25th 53-56 2019年3月

    ISSN:2188-4420

  171. Cross-Corpora Evaluation and Analysis of Grammatical Error Correction Models — Is Single-Corpus Evaluation Enough? 査読有り

    Masato Mita, Tomoya Mizumoto, Masahiro Kaneko, Ryo Nagata, Kentaro Inui

    in Proceedings of the 17th Annual Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies (NAACL-HLT 2019) 1309-1314 2019年2月

    DOI: 10.18653/v1/n19-1132  

  172. Select and Attend: Towards Controllable Content Selection in Text Generation.

    Xiaoyu Shen 0001, Jun Suzuki, Kentaro Inui, Hui Su, Dietrich Klakow, Satoshi Sekine

    CoRR abs/1909.04453 2019年

  173. Riposte! A Large Corpus of Counter-Arguments.

    Paul Reisert, Benjamin Heinzerling, Naoya Inoue, Shun Kiyono, Kentaro Inui

    CoRR abs/1910.03246 2019年

  174. Final Report of the NTCIR-14 QA Lab-PoliInfo Task.

    Yasutomo Kimura, Hideyuki Shibuki, Hokuto Ototake, Yuzu Uchida, Keiichi Takamaru, Kotaro Sakamoto, Madoka Ishioroshi, Teruko Mitamura, Noriko Kando, Tatsunori Mori, Harumichi Yuasa, Satoshi Sekine, Kentaro Inui

    122-135 2019年

    DOI: 10.1007/978-3-030-36805-0_10  

  175. HELP: A Dataset for Identifying Shortcomings of Neural Models in Monotonicity Reasoning.

    Hitomi Yanaka, Koji Mineshima, Daisuke Bekki, Kentaro Inui, Satoshi Sekine, Lasha Abzianidze, Johan Bos

    CoRR abs/1904.12166 2019年

  176. Cross-Corpora Evaluation and Analysis of Grammatical Error Correction Models - Is Single-Corpus Evaluation Enough?

    Masato Mita, Tomoya Mizumoto, Masahiro Kaneko, Ryo Nagata, Kentaro Inui

    CoRR abs/1904.02927 2019年

  177. Stance Detection Attending External Knowledge from Wikipedia. 査読有り

    Kazuaki Hanawa, Akira Sasaki, Naoaki Okazaki, Kentaro Inui

    Journal of Information Processing, Vol. 27 499-506 2019年

  178. クラウドソーシングによる単調推論データセットの構築 査読有り

    谷中 瞳, 戸次 大介, 峯島 宏次, 関根 聡, 乾 健太郎

    人工知能学会全国大会論文集 2019 (0) 2L3J902-2L3J902 2019年

    出版者・発行元:一般社団法人 人工知能学会

    DOI: 10.11517/pjsai.JSAI2019.0_2L3J902  

    詳細を見る 詳細を閉じる

    <p>近年、クラウドソーシングを活用して大規模な言語資源を低コストで構築できるようになった。含意関係認識の研究においても例外ではなく、クラウドソーシングによって構築したデータセットから、ニューラルネットワークを用いて文の意味を学習し、含意関係認識問題を解く手法が活発に研究されている。しかし、クラウドソーシングによって論理的推論を扱う問題を設計することは容易ではなく、既存のデータセットにはそのような問題が含まれていないため、ニューラル含意関係認識モデルが文の論理的な意味を学習できているのかは自明ではない。そこで本研究では、形式意味論の知見とクラウドソーシングの手法を組み合わせて、単調推論のデータセットを構築する手法を提案する。また、現時点でのデータセット構築結果についても報告する。データを分析した結果、論理的に妥当な推論と、語用論的に人間が正しいと判断する推論との間にギャップがあるケースが観察された。</p>

  179. Mixture of Expert/Imitator Networks: Scalable Semi-supervised Learning Framework. 査読有り

    Shun Kiyono, Jun Suzuki, Kentaro Inui

    Proceedings of The Thirty-Third AAAI Conference on Artificial Intelligence (AAAI-19) 4073-4081 2019年1月

    出版者・発行元:None

    DOI: 10.1609/aaai.v33i01.33014073  

  180. 高齢者と家族とのコミュニケーションの質の向上に向けて: 高齢者のQuality of Life 表出発話の分析. 査読有り

    徳久良子, 寺嶌立太, 乾健太郎

    情報処理学会論文誌 2019年

  181. Improving Scientific Relation Classification with Task Specific Supersense. 査読有り

    Qin Dai, Naoya Inoue, Paul Reisert, Kentaro Inui

    Proceedings of the 32nd Pacific Asia Conference on Language, Information and Computing (PACLIC32) 2018年12月

  182. Reducing Odd Generation from Neural Headline Generation. 査読有り

    Shun Kiyono, Sho Takase, Jun Suzuki, Naoaki Okazaki, Kentaro Inui, Masaaki Nagata

    Proceedings of the 32nd Pacific Asia Conference on Language, Information and Computation (PACLIC32) 2018年12月

    出版者・発行元:None

  183. Suspicious News Detection Using Micro Blog Text. 査読有り

    Tsubasa Tagami, Hiroki Ouchi, Hiroki Asano, Kazuaki Hanawa, Kaori Uchiyama, Kaito Suzuki, Kentaro Inui, Atsushi Komiya, Atsuo Fujimura, Ryo Yamashita, Hitofumi Yanai, Akinori Machino

    Proceedings of the 32nd Pacific Asia Conference on Language, Information and Computation (PACLIC 32) 2018年12月

  184. Multi-dialect Neural Machine Translation and Dialectometry. 査読有り

    Kaori Abe, Yuichiroh Matsubayashi, Naoaki Okazaki, Kentaro Inui

    Proceedings of the 32nd Pacific Asia Conference on Language, Information and Computation (PACLIC 32) 2018年12月

    出版者・発行元:None

  185. 文法性・流暢性・意味保存性に基づく文法誤り訂正の参照無し評価. 査読有り

    浅野広樹, 水本智也, 乾健太郎

    自然言語処理 Vol.25 (No.5) 555-576 2018年12月

  186. Leveraging Unannotated Texts for Scientific Relation Extraction. 査読有り

    Qin Dai, Naoya Inoue, Paul Reisert, Kentaro Inui

    IEICE Transactions on Information and Systems 2018年12月

  187. 単語の極性を埋め込んだ分散表現.? 査読有り

    中村拓, 田然, 乾健太郎

    第21回情報論的学習理論ワークショップ (IBIS) 2018年11月

  188. Zunkobot: 複数の知識モジュールを統合した雑談対話システム. 査読有り

    阿部香央莉, 佐藤志貴, 佐藤拓海, 藤井諒, 松田耕史, 鈴木正敏, 山口健史, 赤間怜奈, 大内啓樹, 鈴木潤, 乾健太郎

    第84回 人工知能学会 言語・音声理解と対話処理研究会(SLUD)第9回対話システムシンポジウム 84th 2018年11月

    ISSN:0918-5682

  189. Quality of Life 情報の伝達補助を目的とする対話応答生成. 査読有り

    赤間怜奈, 徳久良子, 乾健太郎

    第84回 人工知能学会 言語・音声理解と対話処理研究会(SLUD)第9回対話システムシンポジウム 2018年11月

  190. An FAQ Search Training Method based on Automatically Generated Questions. 査読有り

    Takuya Makino, Tomoya Noro, Hiyori Yoshikawa, Tomoya Iwakura, Satoshi Sekine, Kentaro Inui

    The 14th Asia Information Retrieval Societies Conference (AIRS 2018) 67-73 2018年11月

    出版者・発行元:None

    DOI: 10.1007/978-3-030-03520-4_7  

  191. What Makes Reading Comprehension Questions Easier?. 査読有り

    Saku Sugawara, Kentaro Inui, Satoshi Sekine, Akiko Aizawa

    Proceedings of the 2018 Conference on Empirical Methods in Natural Language Processing (EMNLP 2018) 4208-4219 2018年11月

    DOI: 10.18653/v1/d18-1453  

  192. Pointwise HSIC: A Linear-Time Kernelized Co-occurrence Norm for Sparse Linguistic Expressions. 査読有り

    Sho Yokoi, Sosuke Kobayashi, Kenji Fukumizu, Jun Suzuki, Kentaro Inui

    Proceedings of the 2018 Conference on Empirical Methods in Natural Language Processing (EMNLP 2018) 1763-1775 2018年11月

    DOI: 10.18653/v1/d18-1203  

  193. Feasible Annotation Scheme for Capturing Policy Argument Reasoning using Argument Templates. 査読有り

    Paul Reisert, Naoya Inoue, Tatsuki Kuribayashi, Kentaro Inui

    Proceedings of The 5th Workshop on Argument Mining 79-89 2018年11月

    出版者・発行元:None

    DOI: 10.18653/v1/w18-5210  

  194. Scientific Knowledge Acquisition via the Interaction between Relation Extraction and Knowledge Graph Completion. 査読有り

    Qin Dai, Naoya Inoue, Paul Reisert, Kentaro Inui

    Proceedings of Third International Workshop on SCIentific DOCument Analysis (SCIDOCA) 2018年11月

  195. Investigating the Challenges of Temporal Relation Extraction from Clinical Text. 査読有り

    Diana Galvan, Naoaki Okazaki, Koji Matsuda, Kentaro Inui

    Proceedings of the Ninth International Workshop on Health Text Mining and Information Analysis (LOUHI 2018) 2018年10月

  196. Unsupervised Token-wise Alignment to Improve Interpretation of Encoder-Decoder Models. 査読有り

    Shun Kiyono, Sho Takase, Jun Suzuki, Naoaki Okazaki, Kentaro Inui, Masaaki Nagata

    Proceedings of the 2018 EMNLP Workshop BlackboxNLP: Analyzing and Interpreting Neural Networks for NLP 74-81 2018年10月

    DOI: 10.18653/v1/w18-5410  

  197. 情報科学論文からの技術の利点・欠点のマイニングに向けて. 査読有り

    白井穂乃, 井之上直也, 乾健太郎

    第13回NLP若手の会 シンポジウム (YANS) 2018年8月

  198. 文法誤り訂正のための教師なし解釈性機構. 査読有り

    三田雅人, 水本智也, 大内啓樹, 永田亮, 乾健太郎

    第13回NLP若手の会 シンポジウム (YANS) 2018年8月

  199. 項目点とその根拠を提示する国語記述式答案自動採点システム. 査読有り

    水本智也, 磯部順子, 乾健太郎

    第13回NLP若手の会 シンポジウム (YANS) 2018年8月

  200. サブワードに基づく単語ベクトルの再構築. 査読有り

    佐々木翔大, 鈴木潤, 乾健太郎

    第13回NLP若手の会 シンポジウム (YANS) 2018年8月

  201. Distance-Free Modeling of Multi-Predicate Interactions in End-to-End Japanese Predicate-Argument Structure Analysis. 査読有り

    Yuichiroh Matsubayashi, Kentaro Inui

    Proceedings of the 27th International Conference on Computational Linguistics (COLING 2018) 94-106 2018年8月

    出版者・発行元:None

  202. Predicting Stances from Social Media Posts using Factorization Machines. 査読有り

    Akira Sasaki, Kazuaki Hanawa, Naoaki Okazaki, Kentaro Inui

    Proceedings of the 27th International Conference on Computational Linguistics (COLING 2018) 3381-3390 2018年8月

  203. Unsupervised Learning of Style-sensitive Word Vectors. 査読有り

    Reina Akama, Kento Watanabe, Sho Yokoi, Sosuke Kobayashi, Kentaro Inui

    Proceedings of the 56th Annual Meeting of the Association for Computational Linguistics (ACL 2018), 572-578 2018年7月

    DOI: 10.18653/v1/P18-2091  

  204. Interpretable and Compositional Relation Learning by Joint Training with an Autoencoder. 査読有り

    Ryo Takahashi, Ran Tian, Kentaro Inui

    Proceedings of the 56th Annual Meeting of the Association for Computational Linguistics (ACL 2018) 2148-2159 2018年7月

  205. 情報科学論文における問題解決手法と評価表現の付与仕様の検討. 査読有り

    白井穂乃, 井之上直也, 乾健太郎

    . 2018年度人工知能学会全国大会(第33回) 2018年6月

  206. ファクトチェックのための要検証記事探索の支援. 査読有り

    内山香, 鈴木海渡, 田上翼, 塙一晃, 乾健太郎, 小宮篤史, 藤村厚夫, 町野明徳, 楊井人文, 山下亮

    . 2018年度人工知能学会全国大会(第32回) 2018年6月

  207. Towards Interpretation as Natural Logic Abduction. 査読有り

    Naoya Inoue, Pontus Stenetorp, Sebastian Riedel, Kentaro Inui

    2018年度人工知能学会全国大会(第32回) 2018年6月

  208. スタイルの類似性を捉えた単語ベクトルの教師なし学習. 査読有り

    赤間怜奈, 渡邉研斗, 横井祥, 小林颯介, 乾健太郎

    2018年度人工知能学会全国大会(第32回) 2018年6月

  209. カーネル法に基づく疎な言語表現のための高速計算可能な共起尺度. 査読有り

    横井祥, 小林颯介, 福水健次, 乾健太郎

    2018年度人工知能学会全国大会(第32回) 2018年6月

  210. Exploring Crowdsourceable Annotation Protocol for Argumentation Schemes. 査読有り

    Naoya Inoue, Paul Reisert, Kentaro Inui

    In IPSJ SIG Technical Reports. Vol.2017-NL-236 1-6 2018年6月

  211. Cross-lingual Learning-to-Rank with Shared Representations. 査読有り

    Shota Sasaki, Shuo Sun, Shigehiko Schamoni, Kevin Duh, Kentaro Inui

    Proceedings of the 16th Annual Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies (NAACL-HLT 2018) 458-463 2018年6月

  212. A Melody-conditioned Lyrics Language Model. 査読有り

    Kento Watanabe, Yuichiroh Matsubayashi, Satoru Fukayama, Masataka Goto, Kentaro Inui, Tomoyasu Nakano

    Proceedings of the 16th Annual Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies (NAACL-HLT 2018) 163-172 2018年6月

    出版者・発行元:None

    DOI: 10.18653/v1/n18-1015  

  213. Modeling storylines in lyrics 査読有り

    Kento Watanabe, Yuichiroh Matsubayashi, Kentaro Inui, Satoru Fukayama, Tomoyasu Nakano, Masataka Goto

    IEICE Transactions on Information and Systems E101D (4) 1167-1179 2018年4月1日

    出版者・発行元:Institute of Electronics, Information and Communication, Engineers, IEICE

    DOI: 10.1587/transinf.2017EDP7188  

    ISSN:1745-1361 0916-8532

    詳細を見る 詳細を閉じる

    This paper addresses the issue of modeling the discourse nature of lyrics and presented the first study aiming at capturing the two common discourse-related notions: storylines and themes. We assume that a storyline is a chain of transitions over topics of segments and a song has at least one entire theme. We then hypothesize that transitions over topics of lyric segments can be captured by a probabilistic topic model which incorporates a distribution over transitions of latent topics and that such a distribution of topic transitions is affected by the theme of lyrics. Aiming to test those hypotheses, this study conducts experiments on the word prediction and segment order prediction tasks exploiting a large-scale corpus of popular music lyrics for both English and Japanese (around 100 thousand songs). The findings we gained from these experiments can be summarized into two respects. First, the models with topic transitions significantly outperformed the model without topic transitions in word prediction. This result indicates that typical storylines included in our lyrics datasets were effectively captured as a probabilistic distribution of transitions over latent topics of segments. Second, themodel incorporating a latent theme variable on top of topic transitions outperformed the models without such variables in both word prediction and segment order prediction. From this result, we can conclude that considering the notion of theme does contribute to the modeling of storylines of lyrics.

  214. Distance-Free Modeling of Multi-Predicate Interactions in End-to-End Japanese Predicate-Argument Structure Analysis. 査読有り

    Yuichiroh Matsubayashi, Kentaro Inui

    Proceedings of the 27th International Conference on Computational Linguistics abs/1806.03869 94-106 2018年

  215. What Makes Reading Comprehension Questions Easier?

    Saku Sugawara, Kentaro Inui, Satoshi Sekine, Akiko Aizawa

    CoRR abs/1808.09384 2018年

  216. ニューラルネットを用いた多方言の翻訳と類型分析

    阿部香央莉, 松林優一郎, 岡崎直観, 乾健太郎, 乾健太郎

    言語処理学会年次大会発表論文集(Web) 24th ROMBINNO.B2‐4 (WEB ONLY) 2018年

    ISSN:2188-4420

  217. 複数の述語間の関係を考慮したEnd‐to‐End日本語述語項構造解析

    松林優一郎, 乾健太郎

    言語処理学会年次大会発表論文集(Web) 24th ROMBUNNO.E1‐3 (WEB ONLY) 2018年

    ISSN:2188-4420

  218. メロディ条件付き歌詞語モデル

    渡邉研斗, 松林優一郎, 深山覚, 乾健太郎, 乾健太郎, 後藤真孝, 中野倫靖

    言語処理学会年次大会発表論文集(Web) 24th ROMBUNNO.D7‐4 (WEB ONLY) 2018年

    ISSN:2188-4420

  219. A joint neural model for fine-grained named entity classification of wikipedia articles 査読有り

    Masatoshi Suzuki, Koji Matsuda, Satoshi Sekine, Naoaki Okazaki, Kentaro Inui

    IEICE Transactions on Information and Systems E101D (1) 73-81 2018年1月1日

    出版者・発行元:Institute of Electronics, Information and Communication, Engineers, IEICE

    DOI: 10.1587/transinf.2017SWP0005  

    ISSN:1745-1361 0916-8532

    詳細を見る 詳細を閉じる

    This paper addresses the task of assigning labels of fine-grained named entity (NE) types to Wikipedia articles. Information of NE types are useful when extracting knowledge of NEs from natural language text. It is common to apply an approach based on supervised machine learning to named entity classification. However, in a setting of classifying into fine-grained types, one big challenge is how to alleviate the data sparseness problem since one may obtain far fewer instances for each fine-grained types. To address this problem, we propose two methods. First, we introduce a multi-task learning framework, in which NE type classifiers are all jointly trained with a neural network. The neural network has a hidden layer, where we expect that effective combinations of input features are learned across different NE types. Second, we propose to extend the input feature set by exploiting the hyperlink structure of Wikipedia. While most of previous studies are focusing on engineering features from the articles’ contents, we observe that the information of the contexts the article is mentioned can also be a useful clue for NE type classification. Concretely, we propose to learn article vectors (i.e. entity embeddings) from Wikipedia’s hyperlink structure using a Skip-gram model. Then we incorporate the learned article vectors into the input feature set for NE type classification. To conduct large-scale practical experiments, we created a new dataset containing over 22,000 manually labeled articles. With the dataset, we empirically show that both of our ideas gained their own statistically significant improvement separately in classification accuracy. Moreover, we show that our proposed methods are particularly effective in labeling infrequent NE types. We’ve made the learned article vectors publicly available. The labeled dataset is available if one contacts the authors.

  220. 文法誤り訂正の文単位評価におけるリファレンスレス手法の評価性能

    浅野広樹, 浅野広樹, 水本智也, 松林優一郎, 乾健太郎, 乾健太郎

    電子情報通信学会技術研究報告 117 (367(NLC2017 35-43)) 13‐20-20 2017年12月13日

    出版者・発行元:電子情報通信学会

    ISSN:0913-5685

  221. A Neural Language Model for Dynamically Representing the Meanings of Unknown Words and Entities in a Discourse 査読有り

    Sosuke Kobayashi, Naoaki Okazaki, Kentaro Inui

    In Proceedings of the 8th International Joint Conference on Natural Language Processing (IJCNLP) 2017年11月

  222. Reference-based Metrics can be Replaced with Reference-less Metrics in Evaluating Grammatical Error Correction Systems 査読有り

    Hiroki Asano, Tomoya Mizumoto, Kentaro Inui

    In Proceedings of the 8th International Joint Conference on Natural Language Processing (IJCNLP) 2017年11月

  223. Generating Stylistically Consistent Dialog Responses with Transfer Learning 査読有り

    Reina Akama, Kazuaki Inada, Naoya Inoue, Sosuke Kobayashi, Kentaro Inui

    In Proceedings of the 8th International Joint Conference on Natural Language Processing (IJCNLP) 2017年11月

  224. Revisiting Design Issues of Local Models for Japanese Predicate-Argument Structure Analysis 査読有り

    Yuichiroh Matsubayashi, Kentaro Inui

    In Proceedings of the 8th International Joint Conference on Natural Language Processing (IJCNLP) 128-133 2017年11月

    出版者・発行元:None

  225. Proofread Sentence Generation as Multi-Task Learning with Edit Operation Prediction 査読有り

    Yuta Hitomi, Hideaki Tamori, Naoaki Okazaki, Kentaro Inui

    In Proceedings of the 8th International Joint Conference on Natural Language Processing (IJCNLP) 2017年11月

  226. A Crowdsourcing Approach for Annotating Causal Relation Instances in Wikipedia 査読有り

    Kazuaki Hanawa, Akira Sasaki, Naoaki Okazaki, Kentaro Inui

    In Proceedings of the 31st Pacific Asia Conference on Language, Information and Computation (PACLIC) 2017年11月

  227. Handling Multiword Expressions in Causality Estimation 査読有り

    Shota Sasaki, Sho Takase, Naoya Inoue, Naoaki Okazaki, Kentaro Inui

    In Proceedings of the 12th International Conference on Computational Semantics (IWCS) 2017年9月

  228. Analyzing the Revision Logs of a Japanese Newspaper for Article Quality Assessment 査読有り

    Hideaki Tamori, Yuta Hitomi, Naoaki Okazaki, Kentaro Inui

    In Proceedings of the 2017 EMNLP Workshop: Natural Language Processing meets Journalism 46-50 2017年9月

  229. Learning Co-Substructures by Kernel Dependence Maximization 査読有り

    Sho Yokoi, Daichi Mochihashi, Ryo Takahashi, Naoaki Okazaki, Kentaro Inui

    In Proceedings of the 26th International Joint Conference on Artificial Intelligence (IJCAI) 3329-3335 2017年8月

    出版者・発行元:None

    DOI: 10.24963/ijcai.2017/465  

  230. Be More Eloquent, Professor ELIZA – Comparison of Utterance Generation Methods for Artificial Second Language Tutor 査読有り

    Taku Nakamura, Rafal Rzepka, Kenji Araki, Kentaro Inui

    In Proceedings of Linguistic and Cognitive Approaches to Dialog Agents (LaCATODA 2017) 2017年8月

  231. The mechanism of additive composition 査読有り

    Ran Tian, Naoaki Okazaki, Kentaro Inui

    MACHINE LEARNING 106 (7) 1083-1130 2017年7月

    出版者・発行元:SPRINGER

    DOI: 10.1007/s10994-017-5634-8  

    ISSN:0885-6125

    eISSN:1573-0565

    詳細を見る 詳細を閉じる

    Additive composition (Foltz et al. in Discourse Process 15:285-307, 1998; Landauer and Dumais in Psychol Rev 104(2):211, 1997; Mitchell and Lapata in Cognit Sci 34(8):1388-1429, 2010) is a widely used method for computing meanings of phrases, which takes the average of vector representations of the constituent words. In this article, we prove an upper bound for the bias of additive composition, which is the first theoretical analysis on compositional frameworks from a machine learning point of view. The bound is written in terms of collocation strength; we prove that the more exclusively two successive words tend to occur together, the more accurate one can guarantee their additive composition as an approximation to the natural phrase vector. Our proof relies on properties of natural language data that are empirically verified, and can be theoretically derived from an assumption that the data is generated from a Hierarchical Pitman-Yor Process. The theory endorses additive composition as a reasonable operation for calculating meanings of phrases, and suggests ways to improve additive compositionality, including: transforming entries of distributional word vectors by a function that meets a specific condition, constructing a novel type of vector representations to make additive composition sensitive to word order, and utilizing singular value decomposition to train word vectors.

  232. メロディと歌詞の相関に基づく自動歌詞生成

    渡邉研斗, 松林優一郎, 深山覚, 中野倫靖, 後藤真孝, 乾健太郎

    情報処理学会研究報告(Web) 2017 (SLP-116) Vol.2017‐SLP‐116,No.16,1‐12 (WEB ONLY) 2017年5月8日

  233. Neural Architectures for Fine-grained Entity Type Classification 査読有り

    Sonse Shimaoka, Pontus Stenetorp, Kentaro Inui, Sebastian Riedel

    In Proceedings of the 15th Conference of the European Chapter of the Association for Computational Linguistics (EACL) 2017年4月

  234. Monitoring Geographical Entities with Temporal Awareness in Tweets 査読有り

    Koji Matsuda, Mizuki Sango, Naoaki Okazaki, Kentaro Inui

    In Proceedings of 18th International Conference on Computational Linguistics and Intelligent Text Processing (CICLING) 2017年4月

  235. Designing a Task for Recognizing Argumentation Logic in Argumentative Texts 査読有り

    Paul Reisert, Naoya Inoue, Naoaki Okazaki, Kentaro Inui

    In Proceedings of 18th International Conference on Computational Linguistics and Intelligent Text Processing (CICLING) 2017年4月

  236. Identifying and Ranking Relevant Claims for Decision Support 査読有り

    Paul Reisert, Naoya Inoue, Naoaki Okazaki, Kentaro Inui

    In Proceedings of 18th International Conference on Computational Linguistics and Intelligent Text Processing (CICLING), 2017年4月

  237. In Proceedings of 18th International Conference on Computational Linguistics and Intelligent Text Processing (CICLING) 招待有り 査読有り

    Paul Reisert, Naoya Inoue, Naoaki Okazaki, Kentaro Inui

    Designing a Task for Recognizing Argumentation Logic in Argumentative Texts 2017年4月

  238. LyriSys: An Interactive Support System for Writing Lyrics Based on Topic Transition 査読有り

    Kento Watanabe, Yuichiroh Matsubayashi, Kentaro Inui, Tomoyasu Nakano, Satoru Fukayama, Masataka Goto

    In Proceedings of the 22nd Annual Meeting of the Intelligent User Interfaces Community (IUI) 559-563 2017年3月

    出版者・発行元:None

    DOI: 10.1145/3025171.3025194  

  239. Disaster Information System Using Natural Language Processing (Special Issue on "Urban Resilience" for Mega Earthquake Disasters(Part 2)) 査読有り

    Kosaka Naoko, Koyama Akira, Kokogawa Tomohiro, Maeda Yuji, Koumoto Hiroko, Suzuki Shingo, Yamaguchi Kenshi, Inui Kentaro

    Journal of disaster research 12 (1) 67-78 2017年2月

    出版者・発行元:Fuji Technology Press

    DOI: 10.20965/jdr.2017.p0067  

    ISSN:1881-2473

  240. Revisiting the Design Issues of Local Models for Japanese Predicate-Argument Structure Analysis.

    Yuichiroh Matsubayashi, Kentaro Inui

    CoRR abs/1710.04437 2017年

  241. A Corpus of Deep Argumentative Structures as an Explanation to Argumentative Relations. 査読有り

    Paul Reisert, Naoya Inoue, Naoaki Okazaki, Kentaro Inui

    CoRR abs/1712.02480 2017年

  242. Leveraging Document-Specific Information for Classifying Relations in Scientific Articles. 査読有り

    Qin Dai, Naoya Inoue, Paul Reisert, Kentaro Inui

    New Frontiers in Artificial Intelligence - JSAI-isAI Workshops, JURISIN, SKL, AI-Biz, LENLS, AAA, SCIDOCA, kNeXI, Tsukuba, Tokyo, Japan, November 13-15, 2017, Revised Selected Papers 355-370 2017年

    出版者・発行元:Springer

    DOI: 10.1007/978-3-319-93794-6_26  

  243. ニューラルネットワークによる日本語述語項構造解析の素性の汎化

    松林優一郎, 乾健太郎

    言語処理学会年次大会発表論文集(Web) 23rd ROMBUNNO.B4‐5 (WEB ONLY) 2017年

    ISSN:2188-4420

  244. Source-side Prediction for Neural Headline Generation. 査読有り

    Shun Kiyono, Sho Takase, Jun Suzuki, Naoaki Okazaki, Kentaro Inui, Masaaki Nagata

    CoRR abs/1712.08302 2017年

  245. Supervised approaches for Japanese Wikification 査読有り

    Shuangshuang Zhou, Naoaki Okazaki, Koji Matsuda, Ran Tian, Kentaro Inui

    Journal of Information Processing 25 341-350 2017年

    出版者・発行元:Information Processing Society of Japan

    DOI: 10.2197/ipsjjip.25.341  

    ISSN:1882-6652 0387-5806

    詳細を見る 詳細を閉じる

    Wikification is the task of connecting mentions in texts to entities in a large-scale knowledge base, Wikipedia. In this paper, we present a pipeline system for Japanese Wikification that consists of two components, namely candidate generation and candidate ranking. We investigate several techniques for each component, using a recently developed Japanese Wikification corpus. For candidate generation, we find that a name dictionary using anchor texts of Wikipedia is more effective than other methods based on similarity of surface forms. For candidate ranking, we verify that a set of features used in English Wikification is effective in Japanese Wikification as well. In addition, by using a corpus that links mentions to Japanese Wikipedia entries instead of to English Wikipedia entries, we are able to acquire rich contextual information from Japanese Wikipedia articles, which leads to improvements for Japanese mention disambiguation. We take this advantage by exploring several embedding models that encode context information of Wikipedia entities. The experimental results demonstrate that they improve candidate ranking. We also report the effect of each feature in detail. To sum, our system achieves 81.60% accuracy, significantly outperforming the previous work.

  246. Annotating Geographical Entities on Microblog Text 査読有り

    Koji Matsuda, Akira Sasaki, Naoaki Okazaki, Kentato Inui

    IPSJ Transactions on Databases (TOD) 2017年

  247. 関係パタンの分散表現の計算 査読有り

    高瀬翔, 岡崎直観, 乾健太郎

    人工知能学会論文誌 32 (4) 1-11 2017年

    出版者・発行元:None

    DOI: 10.1527/tjsai.D-G96  

    ISSN:1346-8030 1346-0714

  248. Geographical entity annotated corpus of Japanese microblogs 査読有り

    Koji Matsuda, Akira Sasaki, Naoaki Okazaki, Kentaro Inui

    Journal of Information Processing 25 121-130 2017年

    出版者・発行元:Information Processing Society of Japan

    DOI: 10.2197/ipsjjip.25.121  

    ISSN:1882-6652 0387-5806

    詳細を見る 詳細を閉じる

    This paper addresses the issues in the task of annotating geographical entities on microblogs and reports the preliminary results of our efforts to annotate Japanese microblog texts. Unlike prior work, we aim at annotating not only geographical location entities but also facility entities, such as stations, restaurants and schools. We discuss (i) how to build a gazetteer of geographical entities with a sufficiently broad coverage, (ii) what types ambiguities that need to be considered, (iii) why the annotator tends to disagree, and (iv) what technical problems should be addressed to automate the task of annotating the geographical entities. All the annotation data and the annotation guidelines are publicly available for research purposes from our web site.

  249. Other topics you may also agree or disagree: Modeling inter-topic preferences using tweets and matrix factorization 査読有り

    Akira Sasaki, Kazuaki Hanawa, Naoaki Okazaki, Kentaro Inui

    ACL 2017 - 55th Annual Meeting of the Association for Computational Linguistics, Proceedings of the Conference (Long Papers) 1 398-408 2017年

    出版者・発行元:Association for Computational Linguistics (ACL)

    DOI: 10.18653/v1/P17-1037  

    詳細を見る 詳細を閉じる

    We present in this paper our approach for modeling inter-topic preferences of Twitter users: for example, those who agree with the Trans-Pacific Partnership (TPP) also agree with free trade. This kind of knowledge is useful not only for stance detection across multiple topics but also for various real-world applications including public opinion surveys, electoral predictions, electoral campaigns, and online debates. In order to extract users' preferences on Twitter, we design linguistic patterns in which people agree and disagree about specific topics (e.g., "A is completely wrong"). By applying these linguistic patterns to a collection of tweets, we extract statements agreeing and disagreeing with various topics. Inspired by previous work on item recommendation, we formalize the task of modeling inter-topic preferences as matrix factorization: representing users' preferences as a user-topic matrix and mapping both users and topics onto a latent feature space that abstracts the preferences. Our experimental results demonstrate both that our proposed approach is useful in predicting missing preferences of users and that the latent vector representations of topics successfully encode inter-topic preferences.

  250. Traffic State Estimation and Cause and Effect Understanding by Probe Data and Social Media Data 査読有り

    Yusuke Hara, Koji Matsuda, Masao Kuwahara, Naoaki Okazaki, Kentaro Inui

    THE 21ST INTERNATIONAL CONFERENCE OF HONG KONG SOCIETY FOR TRANSPORTATION STUDIES, SMART TRANSPORTATION 2016年12月

  251. Modeling Context-sensitive Selectional Preference with Distributed Representations 査読有り

    Naoya Inoue, Yuichiroh Matsubayashi, Masayuki Ono, Naoaki Okazaki, Kentaro Inui

    In Proceedings of the 26th International Conference on Computational Linguistics (COLING 2016) 2829-2838 2016年12月

    出版者・発行元:None

  252. Modeling Discourse Segments in Lyrics Using Repeated Patterns 査読有り

    Kento Watanabe, Yuichiroh Matsubayashi, Naho Orita, Naoaki Okazaki, Kentaro Inui, Satoru Fukayama, Tomoyasu Nakano, Jordan Smith, Masataka Goto

    In Proceedings of the 26th International Conference on Computational Linguistics (COLING 2016) 1959-1969 2016年12月

    出版者・発行元:None

  253. Learning to Describe E-Commerce Images from Noisy Online Data 査読有り

    Takuya Yashima, Naoaki Okazaki, Kentaro Inui, Kota Yamaguchi, Takayuki Okatani

    Proceedings of the 13th Asian Conference on Computer Vision (ACCV '16) (6 pages) 2016年11月

  254. A Pipeline Japanese Entity Linking System with Embedding Features 査読有り

    Shuangshuang Zhou, Koji Matsuda, Ran Tian, Naoaki Okazaki, Kentaro Inui

    Proceedings of the 30th Pacific Asia Conference on Language, Information and Computation (PACLIC 30) 267-276 2016年10月

  255. Recognizing Open-vocabulary Relations between Objects in Images 査読有り

    Masayasu Muraoka, Sumit Maharjan, Masaki Saito, Kota Yamaguchi, Naoaki Okazaki, Takayuki Okatani, Kentaro Inui

    Proceedings of the 30th Pacific Asia Conference on Language, Information and Computation (PACLIC 30) 239-248 2016年10月

  256. Toward the automatic extraction of knowledge of usable goods 査読有り

    Mei Uemura, Naho Orita, Naoaki Okazaki, Kentaro Inui

    In Proceedings of the 30th Pacific Asia Conference on Language 2016年10月

  257. Neural Joint Learning for Classifying Wikipedia Articles into Fine-grained Named Entity Types 査読有り

    Masatoshi Suzuki, Koji Matsuda, Satoshi Sekine, Naoaki Okazaki, Kentaro Inui

    In Proceedings of the 30th Pacific Asia Conference on Language 2016年10月

  258. Multi-label Classification of Wikipedia Articles into Fine-grained Named Entity Types 査読有り

    Masatoshi Suzuki, Koji Matuda, Satoshi Sekine, Naoaki Okazaki, Kentaro Inui

    IEEE/WIC/ACM International Conference on Web Intelligence 2016年10月

  259. Explaining Potential Risks in Traffic Scenes by Combining Logical Inference and Physics Simulation. 査読有り

    Ryo Takahashi, Naoya Inoue, Yasutaka Kuriya, Sosuke Kobayashi, Kentaro Inui

    International Journal of Machine Learning and Computing (IJMLC) 6 (5) 248-255 2016年10月

  260. Building a Corpus for Japanese Wikification with Fine-Grained Entity Classes 査読有り

    Davaajav Jargalsaikhan, Naoaki Okazaki, Koji Matsuda, Kentaro Inui

    In Proceedings of the ACL-2016 Student Research Workshop 2016年8月

  261. Composing Distributed Representations of Relational Patterns 査読有り

    Sho Takase, Naoaki Okazaki, Kentaro Inui

    In Proceedings of the 54th Annual Meeting of the Association for Computational Linguistics (ACL 2016) 2276-2286 2016年8月

  262. Learning Semantically and Additively Compositional Distributional Representations 査読有り

    Ran Tian, Naoaki Okazaki, Kentaro Inui

    In Proceedings of the 54th Annual Meeting of the Association for Computational Linguistics (ACL 2016) 1277-1287 2016年8月

  263. 機能的なリテラルを含む公理体系における仮説推論の効率化 査読有り

    山本風人, 井之上直也, 乾健太郎

    自然言語処理 23 (3) 267-298 2016年6月

    DOI: 10.5715/jnlp.23.267  

  264. Tohoku at SemEval-2016 Task 6: Feature-based Model versus Convolutional Neural Network for Stance Detection 査読有り

    Yuki Igarashi, Hiroya Komatsu, Sosuke Kobayashi, Naoaki Okazaki, Kentaro Inui

    In Proceedings of the 10th International Workshop on Semantic Evaluation (SemEval) 413-419 2016年6月

  265. An Attentive Neural Architecture for Fine-grained Entity Type Classification 査読有り

    Sonse Shimaoka, Pontus Stenetorp, Kentaro Inui, Sebastian Riedel

    In Proceedings of the 5th Workshop on Automated Knowledge Base Construction 2016年6月

  266. Dynamic Entity Representation with Max-pooling Improves Machine Reading 査読有り

    Sosuke Kobayashi, Ran Tian, Naoaki Okazaki, Kentaro Inui

    In Proceedings of the North American Chapter of the Association for Computational Linguistics and Human Language Technologies (NAACL-HLT 2016) 850-855 2016年6月

  267. Question-Answering with Logic Specific to Video Games 査読有り

    Corentin Dumont, Ran Tian, Kentaro Inui

    In Proceedings of the 10th International Conference on the Language Resources and Evaluation Conference (LREC-2016) 4637-4643 2016年5月

  268. Modeling semantic compositionality of relational patterns 査読有り

    Sho Takase, Naoaki Okazaki, Kentaro Inui

    ENGINEERING APPLICATIONS OF ARTIFICIAL INTELLIGENCE 50 256-264 2016年4月

    出版者・発行元:PERGAMON-ELSEVIER SCIENCE LTD

    DOI: 10.1016/j.engappai.2016.01.027  

    ISSN:0952-1976

    eISSN:1873-6769

    詳細を見る 詳細を閉じる

    Vector representation is a common approach for expressing the meaning of a relational pattern. Most previous work obtained a vector of a relational pattern based on the distribution of its context words (e.g., arguments of the relational pattern), regarding the pattern as a single 'word'. However, this approach suffers from the data sparseness problem, because relational patterns are productive, i.e., produced by combinations of words. To address this problem, we propose a novel method for computing the meaning of a relational pattern based on the semantic compositionality of constituent words. We extend the Skip-gram model (Mikolov et al., 2013) to handle semantic compositions of relational patterns using recursive neural networks. The experimental results show the superiority of the proposed method for modeling the meanings of relational patterns, and demonstrate the contribution of this work to the task of relation extraction. (C) 2016 The Authors. Published by Elsevier Ltd.

  269. 分散表現による文脈情報を用いた選択選好モデル

    大野雅之, 井之上直也, 松林優一郎, 岡崎直観, 乾健太郎

    言語処理学会第22回年次大会 22nd 885-888 2016年3月

    ISSN:2188-4420

  270. Fine-Grained Named Entity Classification with Wikipedia Article Vectors 査読有り

    Masatoshi Suzuki, Koji Matsuda, Satoshi Sekine, Naoaki Okazaki, Kentaro Inui

    2016 IEEE/WIC/ACM INTERNATIONAL CONFERENCE ON WEB INTELLIGENCE (WI 2016) 483-486 2016年

    出版者・発行元:IEEE

    DOI: 10.1109/WI.2016.78  

    詳細を見る 詳細を閉じる

    This paper addresses the task of assigning multiple labels of fine-grained named entity (NE) types to Wikipedia articles. To address the sparseness of the input feature space, which is salient particularly in fine-grained type classification, we propose to learn article vectors (i.e. entity embeddings) from hypertext structure of Wikipedia using a Skip-gram model and incorporate them into the input feature set. To conduct large-scale practical experiments, we created a new dataset containing over 22,000 manually labeled instances. The results of our experiments show that our idea gained statistically significant improvements in classification results.

  271. 因果関係に基づく雑談対話発話生成の試み

    佐藤祥多, 井之上直也, 乾健太郎, 樋口佐和, 渡部生聖

    人工知能学会全国大会論文集(CD-ROM) 30th (0) ROMBUNNO.2O5‐OS‐23b‐1-2O5OS23b1 2016年

    出版者・発行元:一般社団法人 人工知能学会

    ISSN:1347-9881

  272. 感情状態に基づく因果関係推論の一般化

    井之上直也, 岡崎直観, 乾健太郎

    言語処理学会年次大会発表論文集(Web) 22nd A6‐5 (WEB ONLY) 2016年

  273. ストーリー展開と一貫性を同時に考慮した歌詞生成モデル

    渡邉研斗, 松林優一郎, 乾健太郎, 深山覚, 中野倫靖, 後藤真孝

    人工知能学会全国大会論文集(CD-ROM) 30th ROMBUNNO.2K3‐1-2K31 2016年

    出版者・発行元:一般社団法人 人工知能学会

    DOI: 10.11517/pjsai.JSAI2016.0_2K31  

    ISSN:1347-9881

  274. 分散表現に基づく選択選好モデルの文脈化

    大野雅之, 井之上直也, 松林優一郎, 岡崎直観, 乾健太郎

    情報処理学会研究報告. 自然言語処理研究会報告 2016-NL-225 (1) 1-9 2016年1月

  275. Stance Classification by Recognizing Related Events about Targets 査読有り

    Akira Sasaki, Junta Mizuno, Naoaki Okazaki, Kentaro Inui

    2016 IEEE/WIC/ACM INTERNATIONAL CONFERENCE ON WEB INTELLIGENCE (WI 2016) 582-587 2016年

    出版者・発行元:IEEE

    DOI: 10.1109/WI.2016.99  

    詳細を見る 詳細を閉じる

    Recently, many people express their opinions using social networking services such as Twitter and Facebook. Each opinion has a stance related to something such as product, service, and politics. The task of detecting a stance is known as sentiment analysis, reputation mining, and stance detection. A popular approach for stance detection uses sentiment polarity towards a target in a text. This approach is known as targeted sentiment analysis. If a target appears in text, the detecting stance based on targeted sentiment polarity would work well. However, how can we detect stance towards an event? (e.g. "I cannot understand why man can marry only with a woman", "The problem of low birth rate becomes more severe" to the event "Allowing same-sex marriage"). To detect these stances, it is necessary to recognize a situation in which the event occurs or does not occur. To classify texts including these phenomena, we propose a classification method based on machine learning considering PRIOR-SITUATION and EFFECT.

  276. 日本語述語項構造解析タスクにおける項の省略を伴う事例の分析 査読有り

    松林優一郎, 中山周, 乾健太郎

    自然言語処理 22 (5) 433-461 2015年12月

    出版者・発行元:一般社団法人 言語処理学会

    DOI: 10.5715/jnlp.22.433  

    ISSN:1340-7619

    詳細を見る 詳細を閉じる

    本稿では,日本語述語項構造解析における中心的課題である項の省略を伴う事例の精度改善を目指し,現象の特徴を詳細に分析することを試みた.具体的には,文内に照応先が出現する事例(文内ゼロ照応)に対象を絞り,人手による手がかりアノテーションと統語的・機能的な構造を元にした機械的分類の二種類の方法により事例を類型化し,カテゴリ毎の分布と最先端のシステムによる解析精度を示した.分析から,特に照応先と直接係り関係にある述語 O が対象述語 P と項を共有する事例が全体の 58% 存在し,O と P の間の統語的・意味的関係が重要な手がかりであることを数値的に示したほか,手がかりの種類や組み合わせが広い分布を持つこと,各手がかりが独立に確信度を上げる事例だけでなく,局所的な手がかりの連鎖が全体で初めて意味を成す事例が一定数存在することを明らかにした.

  277. 誤り分析に基づく日本語事実性解析の課題抽出 査読有り

    成田和弥, 水野淳太, 上岡裕大, 菅野美和, 乾健太郎

    自然言語処理 22 (5) 392-432 2015年12月

    出版者・発行元:一般社団法人 言語処理学会

    DOI: 10.5715/jnlp.22.397  

    ISSN:1340-7619

    詳細を見る 詳細を閉じる

    事実性は,文中の事象の成否について,著者や登場人物の判断を表す情報である.事実性解析には,機能表現や,文節境界を越えて事実性に影響を与える語とそのスコープなどの 4 種類の問題が含まれており,性能の向上が容易ではない.本研究では,事実性解析の課題分析を行うために,機能表現のみを用いたルールベースの事実性解析器を構築し,1,533 文に含まれる 3,734 事象に適用した結果の誤りを分析した.このとき全ての事象表現について,付随する機能表現に対して人手で意味ラベルを付与した.その結果,主事象の事実性解析については,機能表現の意味ラベルが正しく解析できれば,現在の意味ラベルの体系と本研究で用いた単純な規則だけでも,90%に近い正解率が得られることがわかった.従属事象の事実性解析では,後続する述語やスコープといった従属事象特有の誤りが多く見られた.それらの要素についてさらなる分析を行い,今後の事実性解析の指針を示した.

  278. Study on an online communication and task management system for disaster response utilizing natural language processing 査読有り

    Shingo Suzuki, Kentaro Inui, Kenshi Yamaguchi, Hiroko Koumoto, Naoko Kosaka, Akira Koyama, Yuji Maeda

    Journal of Disaster Research 10 (5) 845-856 2015年10月1日

    出版者・発行元:Fuji Technology Press

    DOI: 10.20965/jdr.2015.p0845  

    ISSN:1883-8030 1881-2473

    詳細を見る 詳細を閉じる

    The development and implementation of online-based disaster management information processing systems advance communication among disaster management communities. Many such communities communicate using general-purpose natural language messaging. Online disaster informationprocessing systems should process such communication for making common operational picture and managing tasks and resources. We are thus developing online disaster information management support systems that use natural language processing. In doing so, we compare conventional paper-based and online-based systems for implementing online-based systems and develop task management support systems that use natural language processing.

  279. 日本語文書内で表現される事象間の時間的な順序関係の推定 査読有り

    稲田和明, 松林優一郎, 乾健太郎

    情報処理学会論文誌 56 (10) 2054-2071 2015年10月

    ISSN:1882-7764

    詳細を見る 詳細を閉じる

    本稿では,文書中の表現間の時間的な順序関係の推定(時間関係認識)について述べる.英語では時間関係認識に関する研究が多数存在するが,日本語では多くなく,特に一般的なドメインの文書を対象とした研究が乏しい.そこで本研究では,日本語における頑健な時間関係認識の確立を目指し,ドメインに依存しない解析手法を分析する.英語では機械学習手法によって成果をあげているが,その知見が日本語でも効果的に働くかは自明ではないため,英語で採用される素性が日本語でも同様に有効かを検証する.また,時間関係認識はいまだ発展途上であることから,その解析精度の向上を目指し,時間表現に対する依存構造と大規模データから獲得した頻度情報を利用した素性を提案する.さらに,各素性の貢献度,学習曲線などの分析や,実際の解析によって生じた誤り事例の調査を行う.これらの結果,我々が提案した素性によって解析精度の向上が見られたが,大規模データの頻度情報を利用した素性は頻度獲得に改良が必要と分かった.また,英語での手法は日本語でもおおむね有効であったが,言語間の文法的な表現方法の差異によって,日本語では品詞の素性が悪影響を与えていた.学習曲線などの分析から現状では学習データ量が不十分であり,さらなるデータ量増加に価値があること,誤り分析から事象の持つテンス・アスペクトといった機能的意味の解析などが,今後の時間関係認識の発展に必要と分かった.Temporal relation classification (TRC) is the task of classifying temporal relations between two expressions in a given text. While TRC has been widely explored in English, it has rarely been explored in Japanese aside from domain-specific models. In this work, we aim to create a TRC system for Japanese texts that works with general domains. To accomplish this, we first developed a system based on previous machine-learning models in English, and we verified that the same features work for Japanese. We then evaluated novel features that exploit dependency contexts and a co-occurrence frequency obtained by a large-scale corpus. In detail, we evaluated the contribution of each feature, learning curve, and accuracy of each temporal relation type, and we also performed an error analysis. The results show that our proposed features effectively worked, but it is necessary to improve the co-occurrence-based feature. The features we investigated for English are also effective in Japanese; however, part of speech did not work well due to their grammatical differences. The learning curve indicated that the number of training data was insufficient. Finally, in our error analysis, we found that detecting tense, aspect, and modality are necessary for further improvement of our system.

  280. Fast and Large-scale Unsupervised Relation Extraction 査読有り

    Sho Takase, Naoaki Okazaki, Kentaro Inui

    In Proceedings of the 29th Pacific Asia Conference on LanguageInformation and Computation (PACLIC-29) 96-105 2015年10月

  281. Reducing Lexical features in Parsing by Word Embeddings 査読有り

    Hiroya Komatsu, Ran Tian, Naoaki Okazaki, Kentaro Inui

    In Proceedings of the 29th Pacific Asia Conference on LanguageInformation and Computation (PACLIC-29) 106-113 2015年10月

  282. Recognizing potential traffic risks through logic-based deep scene understanding 査読有り

    Naoya Inoue, Yasutaka Kuriya, Sosuke Kobayashi, Kentaro Inui

    In Proceedings of the 22nd World Congress on Intelligent Transportation Systems (ITSWC2015) 2015年10月

  283. Semantic Annotation of Japanese Functional Expressions and its Impact on Factuality Analysis 査読有り

    Yudai Kamioka, Kazuya Narita, Junta Mizuno, Miwa Kanno, Kentaro Inui

    In Proceedings of the 9th Linguistic Annotation Workshop (LAW IX 2015) 52-61 2015年6月

  284. Annotating Geographical Entities on Microblog Text 査読有り

    Koji Matsuda, Akira Sasaki, Naoaki Okazaki, Kentaro Inui

    In Proceedings of the 9th Linguistic Annotation Workshop (LAW IX 2015) 85-94 2015年6月

  285. A Computational Approach for Generating Toulmin Model Argumentation 査読有り

    Paul Reisert, Naoya Inoue, Naoaki Okazaki, Kentaro Inui

    In Proceedings of the 2nd Workshop on Argumentation Mining 45-55 2015年6月

  286. Learning Sentence Ordering for Opinion Generation of Debate 査読有り

    Toshihiko Yanase, Toshinori Miyoshi, Kohsuke Yanai, Misa Sato, Makoto Iwayama, Yoshiki Niwa, Paul Reisert, Kentaro Inui

    In Proceedings of the 2nd Workshop on Argumentation Mining 94-103 2015年6月

  287. Boosting the Efficiency of First-order Abductive Reasoning Using Pre-estimated Relatedness between Predicates 査読有り

    Kazeto Yamamoto, Naoya Inoue, Kentaro Inui, Yuki Arase, Jun'ichi Tsujii

    International Journal of Machine Learning and Computing 5 (2) 114-120 2015年4月

  288. 日本語述語項構造解析のための統語パターン分析

    中山周, 松林優一郎, 乾健太郎

    言語処理学会第21回年次大会 21st 1028-1031 2015年3月

    ISSN:2188-4420

  289. 大規模歌詞データからの潜在的トピック遷移のモデル化

    渡邉研斗, 松林優一郎, 乾健太郎, 後藤真孝

    情報処理学会 第77回全国大会予稿集 371-372 2015年3月

  290. 日本語文内で表現されるイベント間の時間的な順序関係の認識.

    稲田和明, 松林優一郎, 乾健太郎

    情報処理学会 第77回全国大会予稿集 225-226 2015年3月

  291. 日本語書き言葉を対象とした人間の自然な省略検出の分析 査読有り

    飯田龍, 橋本力, 鳥澤健太郎, 黒橋禎夫, 乾健太郎, 宮尾祐介, 柴田知秀, 笹野遼平

    言語処理学会 第21回年次大会 発表論文集 565-568 2015年3月

  292. LyriSys:歌詞の大局的構造に基づいた作詞支援インタフェース

    渡邉研斗, 松林優一郎, 乾健太郎, 中野倫靖, 深山覚, 後藤真孝

    日本ソフトウェア科学会研究会資料シリーズ(Web) (76) ROMBUNNO.1‐R10 (WEB ONLY) 2015年

    ISSN:1341-870X

  293. Recognizing Complex Negation on Twitter. 査読有り

    Junta Mizuno, Canasai Kruengkrai, Kiyonori Ohtake, Chikara Hashimoto, Kentaro Torisawa, Julien Kloetzer, Kentaro Inui

    In Proceedings of The 29th Pacific Asia Conference on LanguageInformation and Computation (PACLIC-29) 544-552 2015年

    出版者・発行元:ACL

  294. Computer-Assisted Databasing of Disaster Management Information Through Natural Language Processing 査読有り

    Inui Kentaro, Watanabe Yotaro, Yamaguchi Kenshi, Suzuki Shingo, Koumoto Hiroko, Kosaka Naoko, Koyama Akira, Kokogawa Tomohiro, Maeda Yuji

    Journal of Disaster Research 10 (5) 830-844 2015年

    出版者・発行元:富士技術出版株式会社

    DOI: 10.20965/jdr.2015.p0830  

    ISSN:1881-2473

    詳細を見る 詳細を閉じる

    <p>During times of disaster, local government departments and divisions need to communicate a broad range of information for disaster management to share the understating of the changing situation. This paper addresses the issues of how to effectively use a computer database system to communicate disaster management information and how to apply natural language processing technology to reduce the human labor for databasing a vast amount of information. The database schema was designed based on analyzing a collection of real-life disaster management information and the specifications of existing standardized systems. Our data analysis reveals that our database schema sufficiently covers the information exchanged in a local government during the Great East Earthquake. Our prototype system is designed so as to allow local governments to introduce it at a low cost: (i) the system's user interface facilitates the operations for databasing given information, (ii) the system can be easily customized to each local municipality by simply replacing the dictionary and the sample data for training the system, and (iii) the system can be automatically adapted to each local municipality or each disaster incident through its capability of automatic learning from the user's corrections to the system's language processing outputs.</p>

  295. Modeling Structural Topic Transitions for Automatic Lyrics Generation 査読有り

    Kento Watanabe, Yuichiroh Matsubayashi, Kentaro Inui, Masataka Goto

    In Proceedings of the 28th Pacific Asia Conference on LanguageInformation and Computing (PACLIC) 422-431 2014年12月

    出版者・発行元:None

  296. Finding The Best Model Among Representative Compositional Models 査読有り

    Masayasu Muraoka, Sonse Shimaoka, Kazeto Yamamoto, Yotaro Watanabe, Naoaki Okazaki, Kentaro Inui

    In Proceedings of the 28th Pacific Asia Conference on LanguageInformation and Computing (PACLIC) 65-74 2014年12月

  297. Using a Broad Range of Linguistic Features in Entity Discovery and Linking 査読有り

    Shuangshuang Zhou, Canasai Kruengkrai, Naoaki Okazaki, Kentaro Inui

    In Proceedings of Text Analysis Conference 2014 (TAC 2014) 2014年11月

  298. 実用的な自動歌詞生成に向けて

    渡邉研斗, 松林優一郎, 乾健太郎, 後藤真孝

    第9回NLP若手の会 シンポジウム 2014年9月

  299. A Corpus Study for Identifying Evidence on Microblogs 査読有り

    Paul Reisert, Junta Mizuno, Miwa Kanno, Naoaki Okazaki, Kentaro Inui

    In Proceedings of the 8th Linguistic Annotation Workshop (LAW VIII 2014) 70-74 2014年8月

  300. 日本語文章に対する述語項構造アノテーション仕様の考察 査読有り

    松林優一郎, 飯田龍, 笹野遼平, 横野光, 松吉俊, 藤田篤, 宮尾祐介, 乾健太郎

    自然言語処理 21 (2) 333-378 2014年4月

    出版者・発行元:一般社団法人 言語処理学会

    DOI: 10.5715/jnlp.21.333  

    ISSN:1340-7619

    詳細を見る 詳細を閉じる

    日本語の述語項構造アノテーションコーパスは,これまでにいくつかの研究によって整備され,その結果,日本語の述語項構造解析の研究は飛躍的にその成果を伸ばした.一方で,既存のコーパスのアノテーション作業者間一致率やアノテーション結果の定性的な分析をふまえると,ラベル付与に用いる作業用のガイドラインには未だ改善の余地が大きいと言える.本論文では,より洗練された述語項構造アノテーションのガイドラインを作成することを目的とし,NAIST テキストコーパス (NTC), 京都大学テキストコーパス (KTC) のアノテーションガイドラインと実際のラベル付与例を参考に,これらのコーパスの仕様策定,仕様準拠のアノテーションに関わった研究者・アノテータ,仕様の改善に関心のある研究者らの考察をもとにガイドライン策定上の論点をまとめ,現状の問題点や,それらに対する改善策について議論・整理した結果を報告する.また,アノテーションガイドラインを継続的に改善可能とするための方法論についても議論する.

  301. 大局的な構造を考慮した歌詞自動生成システムの提案

    渡邉研斗, 松林優一郎, 乾健太郎, 後藤真孝

    言語処理学会第20回年次大会 20th 694-697 2014年3月

    ISSN:2188-4420

  302. 統計的日本語述語項構造解析のための素性設計再考

    松林優一郎, 乾健太郎

    言語処理学会第20回年次大会 360-363 2014年3月

  303. ILP-based Inference for Cost-based Abduction on First-order Predicate Logic

    Inoue Naoya, Inui Kentaro

    Information and Media Technologies 9 (1) 83-110 2014年

    出版者・発行元:Information and Media Technologies Editorial Board

    DOI: 10.11185/imt.9.83  

    ISSN:1881-0896

    詳細を見る 詳細を閉じる

    Abduction is desirable for many natural language processing (NLP) tasks. While recent advances in large-scale knowledge acquisition warrant applying abduction with large knowledge bases to real-life NLP problems, as of yet, no existing approach to abduction has achieved the efficiency necessary to be a practical solution for large-scale reasoning on real-life problems. In this paper, we propose an efficient solution for large-scale abduction. The contributions of our study are as follows: (i) we propose an efficient method of cost-based abduction in first-order predicate logic that avoids computationally expensive grounding procedures; (ii) we formulate the best-explanation search problem as an integer linear programming optimization problem, making our approach extensible; (iii) we show how cutting plane inference, which is an iterative optimization strategy developed in operations research, can be applied to make abduction in first-order logic tractable; and (iv) the abductive inference engine presented in this paper is made publicly available.

  304. Weighted Abduction for Discourse Processing Based on Integer Linear Programming 査読有り

    Naoya Inoue, Ekaterina Ovchinnikova, Kentaro Inui, Jerry Hobbs

    Plan, Activity, and Intent Recognition: Theory and Practice 33-55 2014年

    出版者・発行元:Elsevier Inc.

    DOI: 10.1016/B978-0-12-398532-3.00002-6  

    詳細を見る 詳細を閉じる

    This chapter explores the logical framework called weighted abduction as applied to solving discourse-processing tasks. Weighted abduction incorporates a cost propagation mechanism allowing us to estimate the likelihood of the obtained abductive proofs. We use a tractable implementation of weighted abduction based on Integer Linear Programming and a large knowledge base generated automatically. We first perform an experiment on plan recognition using the dataset originally developed for Ng and Mooney's system [39]. Then we apply our discourse processing pipeline for predicting whether one text fragment logically entails another one (Recognizing Textual Entailment task). The study we describe is the first attempt to apply tractable inference-based natural language processing on a large scale. © 2014 Elsevier Inc. All rights reserved.

  305. Exploring Linguistic Features for Named Entity Disambiguation. 査読有り

    Shuangshuang Zhou, Canasai Kruengkrai, Naoaki Okazaki, Kentaro Inui

    Int. J. Comput. Linguistics Appl. 5 (2) 49-66 2014年

  306. Mining False Information on Twitter for a Major Disaster Situation 査読有り

    Keita Nabeshima, Junta Mizuno, Naoaki Okazaki, Kentaro Inui

    ACTIVE MEDIA TECHNOLOGY, AMT 2014 8610 96-+ 2014年

    出版者・発行元:SPRINGER-VERLAG BERLIN

    DOI: 10.1007/978-3-319-09912-5_9  

    ISSN:0302-9743

    詳細を見る 詳細を閉じる

    Social networking services (SNS), such as Twitter, disseminate not only useful information, but also false information. Identifying this false information is crucial in order to keep the information on a SNS reliable. The aim of this paper is to develop a method of extracting false information from among a large collection of tweets. We do so by using a set of linguistic patterns formulated to correct false information. More specifically, the proposed method extracts text passages that match specified correction patterns, clusters the passages into topics of false information, and selects a passage that represents each topic of false information. In the experiment we conduct, we build an evaluation set manually, and demonstrate the effectiveness of the proposed method.

  307. An Example-Based Approach to Difficult Pronoun Resolution. 査読有り

    Canasai Kruengkrai, Naoya Inoue, Jun Sugiura, Kentaro Inui

    In Proceedings of the 28th Pacific Asia Conference on LanguageInformation and Computing (PACLIC) 358-367 2014年

    出版者・発行元:The PACLIC 28 Organizing Committee and PACLIC Steering Committee / ACL / Department of Linguistics, Faculty of Arts, Chulalongkorn University

  308. ILP-based Inference for Cost-based Abduction on First-order Predicate Logic 査読有り

    Naoya Inoue, Kentaro Inui

    Journal of Natural Language Processing 20 (5) 629-656 2013年12月

    出版者・発行元:言語処理学会

    DOI: 10.5715/jnlp.20.629  

    ISSN:1340-7619

  309. 日本語述語項構造アノテーションに関わる諸問題の分析

    松林 優一郎, 飯田 龍, 笹野 遼平, 横野 光, 松吉 俊, 藤田 篤, 宮尾 祐介, 乾 健太郎

    情報処理学会 第214回自然言語処理研究会 2013年11月

  310. A Lexicon-based Investigation of Research Issues in Japanese Factuality Analysis 査読有り

    Kazuya Narita, Junta Mizuno, Kentaro Inui

    In Proceedings of the 6th International Joint Conference on Natural Language Processing (IJCNLP 2013) 587-595 2013年10月

  311. NICT Disaster Information Analysis System 査読有り

    Kiyonori Ohtake, Jun Goto, Stijn De Saeger, Kentaro Torisawa, Junta Mizuno, Kentaro Inui

    In Proceedings of the 6th International Joint Conference on Natural Language Processing: System Demonstrations (IJCNLP 2013) 29-32 2013年10月

  312. Extracting and Aggregating False Information from Microblogs 査読有り

    Naoaki Okazaki, Keita Nabeshima, Kento Watanabe, Junta Mizuno, Kentaro Inui

    In Proceedings of the Workshop on Language Processing and Crisis Information 2013 (LPCI 2013) 36-43 2013年10月

  313. Computer-assisted Structuring of Emergency Management Information: A Project Note 査読有り

    Yotaro Watanabe, Kentaro Inui, Shingo Suzuki, Hiroko Koumoto, Mitsuhiro Higashida, Yuji Maeda, Katsumi Iwatasuki

    In Proceedings of the Workshop on Language Processing and Crisis Information 2013 (LPCI 2013) 10-18 2013年10月

  314. Recognizing Implicit Discourse Relations through Abductive Reasoning with Large-scale Lexical Knowledge 査読有り

    Jun Sugiura, Naoya Inoue, Kentaro Inui

    In Proceedings of the 1st Workshop on Natural Language Processing and Automated Reasoning (NLPAR) 76-87 2013年9月

    出版者・発行元:None

  315. Detecting Chronic Critics Based on Sentiment Polarity and User’s Behavior in Social Media 査読有り

    Sho Takase, Akiko Murakami, Miki Enoki, Naoaki Okazaki, Kentaro Inui

    In Proceedings of the 51st Annual Meeting of the Association for Computational Linguistics Proceedings of the Student Research Workshop 110-116 2013年8月

  316. Is a 204 cm Man Tall or Small? Acquisition of Numerical Common Sense from the Web 査読有り

    Katsuma Narisawa, Yotaro Watanabe, Junta Mizuno, Naoaki Okazaki, Kentaro Inui

    In Proceedings of the 51st Annual Meeting of the Association for Computational Linguistics (ACL 2013) 382-391 2013年8月

  317. カテゴリ間の兄弟関係を活用した集合拡張 査読有り

    高瀬翔, 岡崎直観, 乾健太郎

    自然言語処理 20 (2) 273-296 2013年6月

    出版者・発行元:一般社団法人 言語処理学会

    DOI: 10.5715/jnlp.20.273  

    ISSN:1340-7619

    詳細を見る 詳細を閉じる

    集合拡張手法の多くはシードインスタンスだけを手掛かりに新たなインスタンスを取得するものであり,対象が複数のカテゴリであっても,各カテゴリのインスタンスの収集を独立に行う.しかし,複数カテゴリを対象にした集合拡張ではカテゴリ間の関係など,シードインスタンスとは別の事前知識も利用できる.本研究ではこのようなカテゴリ間の関係,特に兄弟関係を事前知識として活用した集合拡張手法を提案する.さらに,Wikipedia から半自動で抽出したインスタンスと兄弟関係を事前知識として実験を行い,兄弟関係が集合拡張に有用であることを示す.

  318. 訂正パターンに基づく誤情報の収集と拡散状況の分析 査読有り

    鍋島啓太, 渡邉研斗, 水野淳太, 岡崎直観, 乾健太郎

    自然言語処理 20 (3) 461-484 2013年6月

    出版者・発行元:一般社団法人 言語処理学会

    DOI: 10.5715/jnlp.20.461  

    ISSN:1340-7619

    詳細を見る 詳細を閉じる

    東日本大震災では,「コスモ石油の爆発で有害物質の雨が降る」などの誤情報の拡散が問題となった.本研究の目的は,東本日大震災後 1 週間の全ツイートから誤情報を網羅的に抽出し,誤情報の拡散と訂正の過程を分析することである.本稿では,誤情報を訂正する表現(以下,訂正パターン)に着目し,誤情報を認識する手法を提案する.具体的には,訂正パターンを人手で整備し,訂正パターンにマッチするツイートを抽出する.次に,収集したツイートを内容の類似性に基づいてクラスタリングし,最後に,その中から誤情報を過不足なく説明する1文を選択する.実験では,誤情報を人手でまとめたウェブサイトを正解データとして,評価を行った.また,誤情報とその訂正情報の拡散状況を,時系列で可視化するシステムを構築した.本システムにより,誤情報の出現・普及,訂正情報の出現・普及の過程を分析できる.

  319. 返信・非公式リツイートに基づくツイート空間の論述構造解析 査読有り

    大和田裕亮, 水野淳太, 岡崎直観, 乾健太郎, 石塚満

    自然言語処理 20 (3) 423-460 2013年6月

    出版者・発行元:一般社団法人 言語処理学会

    DOI: 10.5715/jnlp.20.423  

    ISSN:1340-7619

    詳細を見る 詳細を閉じる

    東日本大震災では安否確認や被災者支援のためにTwitterが活躍したが,一方で多種多様な情報が流通し,混乱を招いた.我々は,情報の信憑性や重要性を評価するには,ツイート空間の論述的な構造を解析・可視化し,情報の「裏」を取ることが大切だと考えている.本稿では,ツイートの返信および非公式リツイート(以下,両者をまとめて返信と略す)に着目し,ツイート間の論述的な関係を認識する手法を提案する.具体的には,返信ツイートによって,投稿者の「同意」「反論」「疑問」などの態度が表明されると考え,これらの態度を推定する分類器を教師有り学習で構築する.評価実験では,返信ツイートで表明される態度の推定性能を報告する.さらに,本手法が直接的に返信関係のないツイート間の論述的な関係の推定にも応用できることを示し,ツイート間の含意関係認識に基づくアプローチとの比較を行う.

  320. THK's Natural Logic-based Compositional Textual Entailment Model at NTCIR-10 RITE-2 査読有り

    Yotaro Watanabe, Junta Mizuno, Kentaro Inui

    In Proceedings of the 10th NTCIR Conference 531-536 2013年6月

  321. TKDDI group at NTCIR10-RITE2: Recognizing Texual Entailment Based on Dependency Structure Alignment 査読有り

    Junta Mizuno, Kentaro Inui, Asuka Sumida, Gen Hattori, Chihiro Ono

    In Proceedings of the 10th NTCIR Conference 553-559 2013年6月

  322. 効率的な推論処理のための日本語文の論理式変換に向けて

    稲田和明, 松林優一郎, 井之上直也, 乾健太郎

    言語処理学会第19回年次大会 19th 608-611 2013年3月

    ISSN:2188-4420

  323. Evidence in Automatic Error Correction Improves Learners’ English Skill 査読有り

    Jiro Umezawa, Junta Mizuno, Naoaki Okazaki, Kentaro Inui

    In Proceedings of the 14th International Conference on Intelligent Text Processing and Computational Linguistics (CICLing 2013) 7817 (2) 559-571 2013年

    DOI: 10.1007/978-3-642-37256-8_46  

    ISSN:0302-9743 1611-3349

  324. Discriminative learning of first-order weighted abduction from partial discourse explanations 査読有り

    Kazeto Yamamoto, Naoya Inoue, Yotaro Watanabe, Naoaki Okazaki, Kentaro Inui

    Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) 7816 (1) 545-558 2013年

    DOI: 10.1007/978-3-642-37247-6_44  

    ISSN:0302-9743 1611-3349

    詳細を見る 詳細を閉じる

    Abduction is inference to the best explanation. Abduction has long been studied in a wide range of contexts and is widely used for modeling artificial intelligence systems, such as diagnostic systems and plan recognition systems. Recent advances in the techniques of automatic world knowledge acquisition and inference technique warrant applying abduction with large knowledge bases to real-life problems. However, less attention has been paid to how to automatically learn score functions, which rank candidate explanations in order of their plausibility. In this paper, we propose a novel approach for learning the score function of first-order logic-based weighted abduction [1] in a supervised manner. Because the manual annotation of abductive explanations (i.e. a set of literals that explains observations) is a time-consuming task in many cases, we propose a framework to learn the score function from partially annotated abductive explanations (i.e. a subset of those literals). More specifically, we assume that we apply abduction to a specific task, where a subset of the best explanation is associated with output labels, and the rest are regarded as hidden variables. We then formulate the learning problem as a task of discriminative structured learning with hidden variables. Our experiments show that our framework successfully reduces the loss in each iteration on a plan recognition dataset. © 2013 Springer-Verlag.

  325. Inducing context gazetteers from encyclopedic databases for named entity recognition 査読有り

    Han-Cheol Cho, Naoaki Okazaki, Kentaro Inui

    Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) 7818 (1) 378-389 2013年

    DOI: 10.1007/978-3-642-37453-1_31  

    ISSN:0302-9743 1611-3349

    詳細を見る 詳細を閉じる

    Named entity recognition (NER) is a fundamental task for mining valuable information from unstructured and semi-structured texts. State-of-the-art NER models mostly employ a supervised machine learning approach that heavily depends on local contexts. However, results of recent research have demonstrated that non-local contexts at the sentence or document level can help advance the improvement of recognition performance. As described in this paper, we propose the use of a context gazetteer, the list of contexts with which entity names can cooccur, as new non-local context information.We build a context gazetteer from an encyclopedic database because manually annotated data are often too few to extract rich and sophisticated context patterns. In addition, dependency path is used as sentence level non-local context to capture more syntactically related contexts to entity mentions than linear context in traditional NER. In the discussion of experimentation used for this study, we build a context gazetteer of gene names and apply it for a biomedical NER task. High confidence context patterns appear in various forms. Some are similar to a predicate-argument structure whereas some are in unexpected forms. The experiment results show that the proposed model using both entity and context gazetteers improves both precision and recall over a strong baseline model, and therefore the usefulness of the context gazetteer. © Springer-Verlag 2013.

  326. 「被災者」ツィートに見る東日本大震災発生1週間の災害対応過程の分析

    佐藤翔輔, 今村文彦, 水野淳太, 岡崎直観, 乾健太郎

    平成24年度東北地域自然災害科学研究集会 2012年12月

  327. Leveraging diverse lexical resources for textual entailment recognition 査読有り

    Yotaro Watanabe, Junta Mizuno, Eric Nichols, Katsuma Narisawa, Keita Nabeshima, Naoaki Okazaki, Kentaro Inui

    ACM Transactions on Asian Language Information Processing 11 (4) 39:1-39:21 2012年12月

    DOI: 10.1145/2382593.2382600  

    ISSN:1530-0226 1558-3430

    詳細を見る 詳細を閉じる

    Since the problem of textual entailment recognition requires capturing semantic relations between diverse expressions of language, linguistic and world knowledge play an important role. In this article, we explore the effectiveness of different types of currently available resources including synonyms, antonyms, hypernym-hyponym relations, and lexical entailment relations for the task of textual entailment recognition. In order to do so, we develop an entailment relation recognition system which utilizes diverse linguistic analyses and resources to align the linguistic units in a pair of texts and identifies entailment relations based on these alignments. We use the Japanese subset of the NTCIR-9 RITE-1 dataset for evaluation and error analysis, conducting ablation testing and evaluation on hand-crafted alignment gold standard data to evaluate the contribution of individual resources. Error analysis shows that existing knowledge sources are effective for RTE, but that their coverage is limited, especially for domain-specific and other low-frequency expressions. To increase alignment coverage on such expressions, we propose a method of alignment inference that uses syntactic and semantic dependency information to identify likely alignments without relying on external resources. Evaluation adding alignment inference to a system using all available knowledge sources shows improvements in both precision and recall of entailment relation recognition. © 2012 ACM 1530-0226/2012/12-ART14 $15.00.

  328. Coreference Resolution with ILP-based Weighted Abduction 査読有り

    Naoya Inoue, Ekaterina Ovchinnikova, Kentaro Inui, Jerry R Hobbs

    In Proceedings of the 24th International Conference on Computational Linguistics (COLING 2012): Technical Papers 1291-1308 2012年12月

    出版者・発行元:None

  329. A Latent Discriminative Model for Compositional Entailment Relation Recognition Using Natural Logic 査読有り

    Yotaro Watanabe, Junta Mizuno, Eric Nichols, Naoaki Okazaki, Kentaro Inui

    In Proceedings of the 24th International Conference on Computational Linguistics (COLING 2012): Technical Papers 2805-2820 2012年12月

  330. Set Expansion using Sibling Relations between Semantic Categories 査読有り

    Sho Takase, Naoaki Okazaki, Kentaro Inui

    In Proceedings of the 26th Pacific Asia Conference on Language Information and Computing 525-534 2012年11月

  331. 説明生成に基づく談話構造解析の課題分析

    杉浦純, 井之上直也, 乾健太郎

    言語処理学会年次大会発表論文集 18th (CD-ROM) ROMBUNNO.D1-6 2012年3月13日

  332. 大規模世界知識を用いた仮説推論による談話解析の課題と対策

    井之上直也, 乾健太郎, OVCHINNIKOVA Ekaterina, HOBBS Jerry R

    言語処理学会年次大会発表論文集 18th (CD-ROM) ROMBUNNO.D1-7 2012年3月13日

  333. Large-scale cost-based abduction in full-fledged first-order predicate logic with cutting plane inference 査読有り

    Naoya Inoue, Kentaro Inui

    Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) 7519 281-293 2012年

    DOI: 10.1007/978-3-642-33353-8_22  

    ISSN:0302-9743 1611-3349

    詳細を見る 詳細を閉じる

    Abduction is inference to the best explanation. Abduction has long been studied intensively in a wide range of contexts, from artificial intelligence research to cognitive science. While recent advances in large-scale knowledge acquisition warrant applying abduction with large knowledge bases to real-life problems, as of yet no existing approach to abduction has achieved both the efficiency and formal expressiveness necessary to be a practical solution for large-scale reasoning on real-life problems. The contributions of our work are the following: (i) we reformulate abduction as an Integer Linear Programming (ILP) optimization problem, providing full support for first-order predicate logic (FOPL) (ii) we employ Cutting Plane Inference, which is an iterative optimization strategy developed in Operations Research for making abductive reasoning in full-fledged FOPL tractable, showing its efficiency on a real-life dataset (iii) the abductive inference engine presented in this paper is made publicly available. © 2012 Springer-Verlag.

  334. Organizing information on the web through agreement-conflict relation classification 査読有り

    Junta Mizuno, Eric Nichols, Yotaro Watanabe, Kentaro Inui

    Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) 7675 126-137 2012年

    DOI: 10.1007/978-3-642-35341-3_11  

    ISSN:0302-9743 1611-3349

    詳細を見る 詳細を閉じる

    The vast amount of information on the Web makes it difficult for users to comprehensively survey the various viewpoints on topics of interest. To help users cope with this information overload, we have developed an Information Organization System that applies state-of-theart technology from Recognizing Textual Entailment to automatically detect Web texts that are relevant to natural language queries and organize them into agreeing and conflicting groups. Users are presented with a bird's-eye-view visualization of the viewpoints on their queries that makes it easier to gain a deeper understanding of an issue. In this paper, we describe the implementation of our Information Organization System and evaluate our system through empirical analysis of the semantic relation recognition system that classifies texts and through a large-scale usability study. The empirical evaluation and usability study both demonstrate the usefulness of our system. User feedback further shows that by exposing our users to differing viewpoints promotes objective thinking and helps to reduce confirmation bias. © Springer-Verlag 2012.

  335. 文間関係認識に基づく賛成・反対意見の俯瞰 査読有り

    水野淳太, 渡邉陽太郎, エリック ニコルズ, 村上浩司, 乾健太郎, 松本裕治

    情報処理学会論文誌 52 (12) 3408-3422 2011年12月

  336. TU Group at NTCIR9-RITE: Leveraging Diverse Lexical Resources for Recognizing Textual Entailment 査読有り

    Yotaro Watanabe, Junta Mizuno, Eric Nichols, Katsuma Narisawa, Keita Nabeshima, Kentaro Inui

    Proceedings of the 9th NTCIR 418-421 2011年12月

  337. ILP-based Reasoning for Weighted Abduction 査読有り

    Naoya Inoue, Kentaro Inui

    In Proceedings of AAAI Workshop on Plan 2011年8月

  338. Knowledge-leveraged Computational Thinking through Natural Language Processing and Statistical Logic (NII Shonan Meeting 2011-4).

    Sadao Kurohashi, Akihiro Yamamoto, Kentaro Inui, Sebastian Riedel 0001

    NII Shonan Meet. Rep. 2011 2011年

  339. Dependency Syntax Analysis Using Grammar Induction and a Lexical Categories Precedence System 査読有り

    Hiram Calvo, Omar J. Gambino, Alexander Gelbukh, Kentaro Inui

    COMPUTATIONAL LINGUISTICS AND INTELLIGENT TEXT PROCESSING, PT I 6608 109-+ 2011年

    出版者・発行元:SPRINGER-VERLAG BERLIN

    DOI: 10.1007/978-3-642-19400-9_9  

    ISSN:0302-9743

    eISSN:1611-3349

    詳細を見る 詳細を閉じる

    The unsupervised approach for syntactic analysis tries to discover the structure of the text using only raw text. In this paper we explore this approach using Grammar Inference Algorithms. Despite of still having room for improvement, our approach tries to minimize the effect of the current limitations of some grammar inductors by adding morphological information before the grammar induction process, and a novel system for converting a shallow parse to dependencies, which reconstructs information about inductor's undiscovered heads by means of a lexical categories precedence system. The performance of our parser, which needs no syntactic tagged resources or rules, trained with a small corpus, is 10% below to that of commercial semi-supervised dependency analyzers for Spanish, and comparable to the state of the art for English.

  340. Co-related Verb Argument Selectional Preferences 査読有り

    Hiram Calvo, Kentaro Inui, Yuji Matsumoto

    COMPUTATIONAL LINGUISTICS AND INTELLIGENT TEXT PROCESSING, PT I 6608 133-+ 2011年

    出版者・発行元:SPRINGER-VERLAG BERLIN

    DOI: 10.1007/978-3-642-19400-9_11  

    ISSN:0302-9743

    eISSN:1611-3349

    詳細を見る 詳細を閉じる

    Learning Selectional Preferences has been approached as a verb and argument problem, or at most as a tri-nary relationship between subject, verb and object. The correlation of all arguments in a sentence, however, has not been extensively studied for sentence plausibility measuring because of the increased number of potential combinations and data sparseness. We propose a unified model for machine learning using SVM (Support Vector Machines) with features based on topic-projected words from a PLSI (Probabilistic Latent Semantic Indexing) Model and PMI (Pointwise Mutual Information) as co-occurrence features, and Word Net top concept projected words as semantic classes. We perform tests using a pseudo-disambiguation task. We found that considering all arguments in a sentence improves the correct identification of plausible sentences with an increase of 10% in recall among other things.

  341. Web information analysis for open-domain decision support: System design and user evaluation 査読有り

    Takuya Kawada, Susumu Akamine, Daisuke Kawahara, Yoshikiyo Kato, Yutaka I. Leon-Suematsu, Kentaro Inui, Sadao Kurohashi, Yutaka Kidawara

    ACM International Conference Proceeding Series 13-18 2011年

    DOI: 10.1145/1964114.1964118  

    詳細を見る 詳細を閉じる

    In this paper, we investigate the effectiveness of the system design of a Web information analysis for open-domain decision support. In order to make decisions, it is required to collect and compare information from various view points. In case of making decisions based on Web information, however, it is difficult to obtain diverse information from variety of sources by using current search engines. Based on this observation, we design a system for supporting open-domain decision making, which analyzes Web information. Among the major design decisions are to focus on two elements, i.e. identifying the source of information and the extraction of informative content, and to organize the two elements so that the user can quickly grasp who is saying what on the Web. The assumption behind such decisions is that information organized in such a way would facilitate proper judgments in the user's decision making process. We conduct users evaluation to verify the effectiveness of our approach. In the result, it is confirmed that our system is superior to current search engine for grasping organized information from different stance of senders and supports the process of decision making, by (i) uncovering biases, (ii) showing various opinions from multiple view points, (iii) revealing information sources. © 2011 ACM.

  342. Web spam detection by exploring densely connected subgraphs 査読有り

    Yutaka I. Leon-Suematsu, Kentaro Inui, Sadao Kurohashi, Yutaka Kidawara

    Proceedings - 2011 IEEE/WIC/ACM International Conference on Web Intelligence, WI 2011 1 124-129 2011年

    DOI: 10.1109/WI-IAT.2011.152  

    詳細を見る 詳細を閉じる

    In this paper, we present a Web spam detection algorithm that relies on link analysis. The method consists of three steps: (1) decomposition of webgraphs in densely connected subgraphs and calculation of the features for each subgraph (2) use of SVM classifiers to identify subgraphs composed of Web spam and (3) propagation of predictions over webgraphs by a biased PageRank algorithm to expand the scope of identification. We performed experiments on a public benchmark. An empirical study of the core structure of webgraphs suggests that highly ranked non-spam hosts can be identified by viewing the coreness of the webgraph elements. © 2011 IEEE.

  343. Recognizing Confinement in Web Texts 査読有り

    Megumi Ohki, Eric Nichols, Suguru Matsuyoshi, Koji Murakami, Junta Mizuno, Shouko Masuda, Kentaro Inui, Yuji Matsumoto

    he 9th International Conference on Computational Semantics (IWCS 2011) 2011年1月

  344. Mining personal experiences and opinions from Web documents 査読有り

    Shuya Abe, Kentaro Inui, Kazuo Hara, Hiraku Morita, Chitose Sao, Megumi Eguchi, Asuka Sumita, Koji Murakami, Suguru Matsuyoshi

    Web Intelligence and Agent Systems 9 (2) 109-121 2011年

    DOI: 10.3233/WIA-2011-0209  

    ISSN:1570-1263

    詳細を見る 詳細を閉じる

    This paper proposes a new UGC-oriented language technology application, which we call experience mining. Experience mining aims at automatically collecting instances of personal experiences as well as opinions from vast amounts of user generated content (UGC) such as weblog and forum posts and storing them in an experience database with semantically rich indices. After discussing the technical issues relating to this new task, we focus on the central problem of factuality analysis, formulate a task definition, and propose a machine learning-based solution. Our empirical evaluation indicates that our factuality analysis defintion is sufficiently well-defined to achieve a high inter-annotator agreement and our Factorial CRF-based model considerably outperforms the baseline. We also present an application system, which currently stores over 50M experience instances extracted from 150M Japanese blog posts with semantic indices and serves an experience search engine for unrestricted users and report on our empirical evaluation of the system's accuracy. © 2011 - IOS Press and the authors. All rights reserved.

  345. Web spam detection by exploring densely connected subgraphs 査読有り

    Yutaka I. Leon-Suematsu, Kentaro Inui, Sadao Kurohashi, Yutaka Kidawara

    Proceedings - 2011 IEEE/WIC/ACM International Conference on Web Intelligence, WI 2011 1 124-129 2011年

    DOI: 10.1109/WI-IAT.2011.152  

    詳細を見る 詳細を閉じる

    In this paper, we present a Web spam detection algorithm that relies on link analysis. The method consists of three steps: (1) decomposition of webgraphs in densely connected subgraphs and calculation of the features for each subgraph (2) use of SVM classifiers to identify subgraphs composed of Web spam and (3) propagation of predictions over webgraphs by a biased PageRank algorithm to expand the scope of identification. We performed experiments on a public benchmark. An empirical study of the core structure of webgraphs suggests that highly ranked non-spam hosts can be identified by viewing the coreness of the webgraph elements. © 2011 IEEE.

  346. Automatic Classification of Semantic Relations between Facts and Opinions. 査読有り

    Koji Murakami, Eric Nichols, Junta Mizuno, Yotaro Watanabe, Hayato Goto, Megumi Ohki, Suguru Matsuyoshi, Kentaro Inui, Yuji Matsumoto

    Proceedings of the Second International Workshop on NLP Challenges in the Information Explosion Era (NLPIX 2010) 2010年8月

  347. Identifying contradictory and contrastive relations between statements to outline Web information on a given topic. 査読有り

    Daisuke Kawahara, Kentaro Inui, Sadao Kurohashi

    Proceedings of the 23rd International Conference on Computational Linguistics (COLING) 2010年8月

  348. A Thesaurus of Predicate-Argument Structure for Japanese Verbs to Deal with Granularity of Verb Meanings 査読有り

    Koichi Takeuchi, Kentaro Inui, Nao Takeuchi, Atsushi Fujita

    Proceedings of the 8th Workshop on Asian Language Resources 1-8 2010年8月

  349. テキスト情報分析のための判断情報アノテーション 査読有り

    松吉俊, 江口萌, 佐尾ちとせ, 村上浩司, 乾健太郎, 松本裕治

    電子情報通信学会論文誌 J93-D (6) 705-713 2010年6月

  350. Dependency tree-based sentiment classification using CRFs with hidden variables. 査読有り

    Tetsuji Nakagawa, Kentaro Inui, Sadao Kurohashi

    Proceedings of the North American Chapter of the Association for Computational Linguistics and Human Language Technologies (NAACL-HLT), 1778-1785 2010年6月

    詳細を見る 詳細を閉じる

    In this paper, we present a dependency treebased method for sentiment classification of Japanese and English subjective sentences using conditional random fields with hidden variables. Subjective sentences often contain words which reverse the sentiment polarities of other words. Therefore, interactions between words need to be considered in sentiment classification, which is difficult to be handled with simple bag-of-words approaches, and the syntactic dependency structures of subjective sentences are exploited in our method. In the method, the sentiment polarity of each dependency subtree in a sentence, which is not observable in training data, is represented by a hidden variable. The polarity of the whole sentence is calculated in consideration of interactions between the hidden variables. Sum-product belief propagation is used for inference. Experimental results of sentiment classification for Japanese and English subjective sentences showed that the method performs better than other methods based on bag-of-features. 1

  351. 述語項構造と照応関係のアノテーション: NAISTテキストコーパス構築の経験から 査読有り

    飯田龍, 小町守, 乾健太郎, 松本裕治

    自然言語処理 17 (2) 25-50 2010年4月

    DOI: 10.5715/jnlp.17.2_25  

  352. Resolving Direct and Indirect Anaphora for Japanese Definite Noun Phrases. 査読有り

    Naoya Inoue, Ryu Iida, Kentaro Inui, Yuji Matsumoto

    Journal of Natural Language Processing 17 (1) 221-246 2010年4月

    DOI: 10.5715/jnlp.17.1_221  

  353. 名詞句の語彙統語パターンを用いた事態性名詞の項構造解析 査読有り

    小町守, 飯田龍, 乾健太郎, 松本裕治

    自然言語処理 17 (1) 141-160 2010年1月

    DOI: 10.5715/jnlp.17.1_141  

  354. 項の共有関係と統語パターンを用いた事態間関係獲得 査読有り

    阿部修也, 乾健太郎, 松本裕治

    自然言語処理 17 (1) 121-140 2010年1月

    DOI: 10.5715/jnlp.17.1_121  

  355. Web ページの情報発信者の同定 査読有り

    加藤義清, 河原大輔, 乾健太郎, 黒橋禎夫, 柴田知秀

    人工知能学会論文誌 25 (1) 90-103 2010年

    出版者・発行元:None

    DOI: 10.1527/tjsai.25.90  

    ISSN:1346-8030 1346-0714

  356. Organizing information on the web to support user judgments on information credibility 査読有り

    Susumu Akamine, Daisuke Kawahara, Yoshikiyo Kato, Tetsuji Nakagawa, Yutaka I. Leon-Suematsu, Takuya Kawada, Kentaro Inui, Sadao Kurohashi, Yutaka Kidawara

    2010 4th International Universal Communication Symposium, IUCS 2010 - Proceedings 123-130 2010年

    DOI: 10.1109/IUCS.2010.5666759  

    詳細を見る 詳細を閉じる

    A vast amount of information and knowledge has been accumulated and circulated on the Web. They provide people with options regarding their daily lives and are starting to have a strong influence on governmental policies and business management. A crucial problem is that information on the Web is not necessarily credible. This paper describes an information analysis system called WISDOM, which assists users in assessing the credibility of information on the Web. WISDOM is to organize information on a given topic through the following three types of analyses: (1) extracting and contrasting opinions and important statements around the points related to the topic, (2) identifying and classifying the information sender of each page and (3) analyzing the appearance of each page, for example, page design and writing style. Our preliminary evaluation indicates the effectiveness of WISDOM and its advantage to Google from the viewpoint of the ability of grasping the difference of information senders and opinions. ©2010 IEEE.

  357. Expertise analysis of information senders of web pages 査読有り

    Yoshikiyo Kato, Kentaro Inui, Sadao Kurohashi

    2010 4th International Universal Communication Symposium, IUCS 2010 - Proceedings 135-140 2010年

    DOI: 10.1109/IUCS.2010.5666231  

    詳細を見る 詳細を閉じる

    In this study, we address the problem of searching experts in an arbitrary topic on the Web. In particular, we propose two methods that analyze the expertise of information senders of Web pages: 1) a method that computes expertise score based on hit count from a search engine (hit count method), and 2) a method that computes expertise score based on the number of documents that are attributed to an information sender (attribution count method). We evaluated both methods using a crawl and search infrastructure which indexes 120 million Japanese Web pages. The results show that the attribution count method outperforms hit count method in terms of precision and processing time. ©2010 IEEE.

  358. Simulator platform that enables social interaction simulation - SIGVerse: SocioIntelliGenesis simulator 査読有り

    Tetsunari Inamura, Tomohiro Shibata, Hideaki Sena, Takashi Hashimoto, Nobuyuki Kawai, Takahiro Miyashita, Yoshiki Sakurai, Masahiro Shimizu, Mihoko Otake, Koh Hosoda, Satoshi Umeda, Kentaro Inui, Yuichiro Yoshikawa

    2010 IEEE/SICE International Symposium on System Integration: SI International 2010 - The 3rd Symposium on System Integration, SII 2010, Proceedings 212-217 2010年

    DOI: 10.1109/SII.2010.5708327  

    詳細を見る 詳細を閉じる

    Understanding mechanisms of intelligence of human beings and animals is one of the most important approaches to develop intelligent robot systems. Since the mechanisms of such real-life intelligent systems are so complex, physical interactions between agents and their environment and the social interactions between agents should be considered. Comprehension and knowledge in many peripheral fields such as cognitive science, developmental psychology, brain science, evolutionary biology, and robotics is also required. Discussions from an interdisciplinary aspect are very important for implementing this approach, but such collaborative research is time-consuming and labor-intensive, and it is difficult to obtain fruitful results from such research because the basis of experiments is very different in each research field. In the social science field, for example, several multi-agent simulation systems have been proposed for modeling factors such as social interactions and language evolution, whereas robotics researchers often use dynamics and sensor simulators. However, there is no integrated system that uses both physical simulations and social communication simulations. Therefore, we developed a simulator environment called SIGVerse that combines dynamics, perception, and communication simulations for synthetic approaches to research into the genesis of social intelligence. In this paper, we introduce SIGVerse, its example application and perspectives. ©2010 IEEE.

  359. Annotating Event Mentions in Text with Modality, Focus, and Source Information 査読有り

    Suguru Matsuyoshi, Megumi Eguchi, Chitose Sao, Koji Murakami, Kentaro Inui, Yuji Matsumoto

    LREC 2010 - SEVENTH INTERNATIONAL CONFERENCE ON LANGUAGE RESOURCES AND EVALUATION 1456-1463 2010年

    出版者・発行元:EUROPEAN LANGUAGE RESOURCES ASSOC-ELRA

    詳細を見る 詳細を閉じる

    Many natural language processing tasks, including information extraction, question answering and recognizing textual entailment, require analysis of the polarity, focus of polarity, tense, aspect, mood and source of the event mentions in a text in addition to its predicate argument structure analysis. We refer to modality, polarity and other associated information as extended modality. In this paper, we propose a new annotation scheme for representing the extended modality of event mentions in a sentence. Our extended modality consists of the following seven components: Source, Time, Conditional, Primary modality type, Actuality, Evaluation and Focus. We reviewed the literature about extended modality in Linguistics and Natural Language Processing (NLP) and defined appropriate labels of each component. In the proposed annotation scheme, information of extended modality of an event mention is summarized at the core predicate of the event mention for immediate use in NLP applications. We also report on the current progress of our manual annotation of a Japanese corpus of about 50,000 event mentions, showing a reasonably high ratio of inter-annotator agreement.

  360. Statement map: Reducing web information credibility noise through opinion classification 査読有り

    Koji Murakami, Eric Nichols, Junta Mizuno, Yotaro Watanabe, Shouko Masuda, Hayato Goto, Megumi Ohki, Chitose Sao, Suguru Matsuyoshi, Kentaro Inui, Yuji Matsumoto

    International Conference on Information and Knowledge Management, Proceedings 59-66 2010年

    DOI: 10.1145/1871840.1871850  

    詳細を見る 詳細を閉じる

    On the Internet, users often encounter noise in the form of spelling errors or unknown words, however, dishonest, unreliable, or biased information also acts as noise that makes it difficult to find credible sources of information. As people come to rely on the Internet for more and more information, reducing this credibility noise grows ever more urgent. The Statement Map project's goal is to help Internet users evaluate the credibility of information sources by mining the Web for a variety of viewpoints on their topics of interest and presenting them to users together with supporting evidence in a way that makes it clear how they are related. In this paper, we show how a Statement Map system can be constructed by combining Information Retrieval (IR) and Natural Language Processing (NLP) technologies, focusing on the task of organizing statements retrieved from the Web by viewpoints. We frame this as a semantic relation classification task, and identify 4 semantic relations: [Agreement], [Conflict], [Confinement], and [Evidence]. The former two relations are identified by measuring semantic similarity through sentence alignment, while the latter two are identified through sentence-internal discourse processing. As a prelude to end-to-end user evaluation of Statement Map, we present a large-scale evaluation of semantic relation classification between user queries and Internet texts in Japanese and conduct detailed error analysis to identify the remaining areas of improvement. © 2010 ACM.

  361. 共起パターンの学習による事態間関係知識の獲得 査読有り

    阿部修也, 乾健太郎, 松本裕治

    自然言語処理 16 (4) 79-82 2009年10月

    DOI: 10.5715/jnlp.16.5_79  

  362. Resolving Direct and Indirect Anaphora for Japanese Definite Noun Phrases. 査読有り

    Naoya Inoue, Ryu Iida, Kentaro Inui, Yuji Matsumoto

    Proceedings of the Conference of the Pacific Association for Computational Linguistics (PACLING), 268-273 2009年9月

  363. Constructing a Scientific Blog Corpus for Information Credibility Analysis. 査読有り

    Eric Nichols, Koji Murakami, Kentaro Inui, Yuji Matsumoto

    Proceedings of the Conference of the Pacific Association for Computational Linguistics (PACLING), 302-307 2009年9月

  364. Capturing Salience with a Trainable Cache Model for Zero-anaphora Resolution. 査読有り

    Ryu Iida, Kentaro Inui, Yuji Matsumoto

    Proceedings of Joint Conference of the 47th Annual Meeting of the Association for Computational Linguistics and the 4th International Joint Conference on Natural Language Processing (ACL-IJCNLP), 647-655 2009年8月

    出版者・発行元:Association for Computational Linguistics

  365. WISDOM: A Web Information Credibility Analysis System. 査読有り

    Susumu Akamine, Daisuke Kawahara, Yoshikiyo Kato, Tetsuji Nakagawa, Kentaro Inui, Sadao Kurohashi, Yutaka Kidawara

    Proceedings of Joint Conference of the 47th Annual Meeting of the Association for Computational Linguistics and the 4th International Joint Conference on Natural Language Processing (ACL-IJCNLP), 1-4 2009年8月

    詳細を見る 詳細を閉じる

    We demonstrate an information credibility analysis system called WISDOM. The purpose of WISDOM is to evaluate the credibility of information available on the Web from multiple viewpoints. WISDOM considers the following to be the source of information credibility: information contents, information senders, and information appearances. We aim at analyzing and organizing these measures on the basis of semantics-oriented natural language processing (NLP) techniques. 1.

  366. Annotating Semantic Relations Combining Facts and Opinions. 査読有り

    Koji Murakami, Shouko Masuda, Suguru Matsuyoshi, Eric Nichols, Kentaro Inui, Yuji Matsumoto

    Proceedings of the Third Linguistic Annotation Workshop (LAW III), 150-153 2009年8月

  367. Webから獲得した感情生起要因コーパスに基づく感情推定 査読有り

    徳久良子, 乾健太郎, 松本裕治

    情報処理学会論文誌 50 (4) 1365-1374 2009年4月

  368. 指定指示・代行指示を区別した指示連体詞の照応解析

    井之上直也, 飯田龍, 乾健太郎, 松本裕治

    言語処理学会年次大会発表論文集 15th 372-375 2009年3月2日

  369. Statement Map: Assisting Information Credibility Analysis by Visualizing Arguments 査読有り

    Koji Murakami, Eric Nichols, Suguru Matsuyoshi, Asuka Sumida, Shouko Masuda, Kentaro Inui, Yuji Matsumoto

    WICOW 09 43-50 2009年

    出版者・発行元:ASSOC COMPUTING MACHINERY

    詳細を見る 詳細を閉じる

    In this paper we introduce STATEMENT MAP, a project designed to help users navigate the vast amounts of information on the internet and come to informed opinions on topics of interest. It does this by mining the Web for a variety of viewpoints and presenting them to users together with supporting evidence in a way that makes it clear how the viewpoints are related. In this paper, we discuss the need to address issues of information credibility on the internet, outline the development of STATEMENT MAP generators for Japanese and English, discuss the technical issues that are being addressed, and report on the construction of the resources necessary to meet the project&apos;s goals.

  370. Identifying Information Sender Configuration of Web Pages 査読有り

    Yoshikiyo Kato, Daisuke Kawahara, Kentaro Inui, Sadao Kurohashi, Tomohide Shibata

    2009 IEEE/WIC/ACM INTERNATIONAL JOINT CONFERENCES ON WEB INTELLIGENCE (WI) AND INTELLIGENT AGENT TECHNOLOGIES (IAT), VOL 1 335-340 2009年

    出版者・発行元:IEEE COMPUTER SOC

    DOI: 10.1109/WI-IAT.2009.59  

    詳細を見る 詳細を閉じる

    The source of a piece of information is a crucial element to consider when judging the credibility of that information. In this paper: we address the task of identifying the information source which is cast as a problem of identifying the information sender configuration (ISC) of a Web page. An information sender of a Web page is an entity which is involved in the publication of the information on the page. An ISC of a Web page describes the information senders of the page and the relationship among them. Information sender extraction is thus a subtask of identifying ISC, and we present a method for extracting information senders from Web pages and offer preliminary evaluation. The ISC provides a basis for deeper analysis of information on the Web.

  371. Development of a large-scale web crawler and search engine infrastructure 査読有り

    Susumu Akamine, Yoshikiyo Kato, Daisuke Kawahara, Keiji Shinzato, Kentaro Inui, Sadao Kurohashi, Yutaka Kidawara

    ACM International Conference Proceeding Series 126-131 2009年

    DOI: 10.1145/1667780.1667805  

    詳細を見る 詳細を閉じる

    This paper reports the ongoing development of a large-scale Web crawler and search engine infrastructure at National Institute of Information and Communications Technology. This infrastructure has the following characteristics: (1) It collects one billion Japanese Web pages while keeping them up-to-date. (2) It selects 100 million pages from among the collected pages and converts them into a standard data format to store the results of morphological analysis, dependency parsing, and synonym augmentation. (3) The selected set of pages is searchable and accessible to the users. (4) The scalability of the system is achieved by using a large-scale cluster machine for distributed data processing. Copyright 2009 ACM.

  372. Summarizing evaluative information on the web for information credibility analysis 査読有り

    Daisuke Kawahara, Tetsuji Nakagawa, Takuya Kawada, Kentaro Inui, Sadao Kurohashi

    ACM International Conference Proceeding Series 187-192 2009年

    DOI: 10.1145/1667780.1667819  

    詳細を見る 詳細を閉じる

    The World Wide Web comprises a wide variety of evaluative information. It consists of positive and negative opinions on innumerable topics from various perspectives, thus proving to be a useful information source for information credibility analysis. To present an informative and at-a-glance summary of any topic that a user of such an analysis system searches for, it is important to summarize many diverse evaluative expressions on the topic. In this paper, we describe a method for summarizing an extensive variety of evaluative expressions that are automatically extracted. Copyright 2009 ACM.

  373. Topic relatedness in evaluative information extraction 査読有り

    Takuya Kawada, Tetsuji Nakagawa, Kentaro Inui, Sadao Kurohashi

    ACM International Conference Proceeding Series 120-125 2009年

    DOI: 10.1145/1667780.1667804  

    詳細を見る 詳細を閉じる

    The task of extracting opinions/evaluations related to a given topic from a large number of documents such as Web documents is crucial for developing an automatic evaluation finding system, which can handle a wide variety of topics as input. In this paper, we discuss the topic relatedness of extracted evaluation through analysis of a corpus we developed. We suggest here that the semantic relationship between the target of each extracted evaluation and a given topic helps in judging topic relatedness. In addition, we point out other factors that are beyond the analysis of topic-target relations for judging the topic relatedness of evaluation. Copyright 2009 ACM.

  374. 1.キーワードサーチを超える情報爆発サーチ : 自然言語処理で価値ある未知をマイニング(パートI:情報爆発時代における新しい基盤技術,<特集>情報爆発時代におけるわくわくするITの創出を目指して) 招待有り

    鳥澤健太郎, 中川裕志, 黒橋禎夫, 乾健太郎, 吉岡真治, 藤井敦, 喜連川優

    情報処理 49 (8) 890-896 2008年8月

    出版者・発行元:一般社団法人情報処理学会

    ISSN:0447-8053

  375. Multiple Purpose Annotation using SLAT - Segment and Link-based Annotation Tool - 査読有り

    Masaki Noguchi, Kenta Miyoshi, Takenobu Tokunaga, Ryu Iida, Mamoru Komachi, Kentaro Inui

    Proceedings of 2nd Linguistic Annotation Workshop, 2008 pp. 61-64 2008年5月

  376. Extracting the author of web pages 査読有り

    Yoshikiyo Kato, Daisuke Kawahara, Kentaro Inui, Sadao Kurohashi, Tomohide Shibata

    International Conference on Information and Knowledge Management, Proceedings 35-41 2008年

    DOI: 10.1145/1458527.1458537  

    詳細を見る 詳細を閉じる

    In this paper, we define the problem of identifying the author of a Web page as a sub-problem of identifying the information sender configuration of a Web page. We propose a method that extracts the author name candidates from a Web page based on linguistic features, and rank the candidates based on local features such as distance from the main content. The evaluation shows that we can achieve more than 75% precision when evaluated with candidates ranked within top five. Copyright 2008 ACM.

  377. Extracting Subjective and Objective Evaluative Expressions from the Web 査読有り

    Tetsuji Nakagawa, Takuya Kawada, Kentaro Inui, Sadao Kurohashi

    PROCEEDINGS OF THE SECOND INTERNATIONAL SYMPOSIUM ON UNIVERSAL COMMUNICATION 251-258 2008年

    出版者・発行元:IEEE COMPUTER SOC

    DOI: 10.1109/ISUC.2008.17  

    詳細を見る 詳細を閉じる

    There are various opinions on the Web, and analyzing them is an important task Although many previous studies focused on analyzing subjective evaluative expressions, objective evaluative expressions which describe positive or negative facts are also informative information. In this paper we study extraction and classification of subjective and objective evaluative expressions on Japanese Web documents using machine learning and evaluative word dictionaries.

  378. Using Appearance Information for Web Information Credibility Analysis 査読有り

    Susumu Akamine, Yoshikiyo Kato, Kentaro Inui, Sadao Kurohashi

    PROCEEDINGS OF THE SECOND INTERNATIONAL SYMPOSIUM ON UNIVERSAL COMMUNICATION 363-365 2008年

    出版者・発行元:IEEE COMPUTER SOC

    DOI: 10.1109/ISUC.2008.80  

    詳細を見る 詳細を閉じる

    We have been developing a Web information analysis system named WISDOM, which supports human judgment of information credibility by analyzing Web pages based on various criteria such as the sender, content, and appearance. This paper describes use of appearance information on WISDOM. Appearance information means document style and superficial characteristic, such as page layout, sentential style, the number of advertisement, and whether or not there is contract address. We use appearance information for not only Web information credibility analysis but also Web page classification. This paper also reports the extraction method of appearance information, implementation on WISDOM, and results of preliminary experiment.

  379. Classifying information sender of web documents 査読有り

    Yoshikiyo Kato, Sadao Kurohashi, Kentaro Inui

    INTERNET RESEARCH 18 (2) 191-203 2008年

    出版者・発行元:EMERALD GROUP PUBLISHING LIMITED

    ISSN:1066-2243

    詳細を見る 詳細を閉じる

    Purpose - To develop a method for classifying information sender of web documents, which constitutes an important part of information credibility analysis. Design/methodology/approach - Machine learning approach was employed. About 2,000 human-annotated web documents were prepared for training and evaluation. The classification model was based on support vector machine, and the features used for the classification included the title and URL of documents, as well as information of the top page. Findings -With relatively small set of features, the proposed method achieved over 50 per cent accuracy. Research limitations/implications - Some of the information sender categories were found to be more difficult to classify. This is due to the subjective nature of the categories, and further refinement of the categories is needed. Practical implications - When combined with opinion/sentiment analysis techniques, information sender classification allows more profound analysis based on interactions between opinions and senders. Such analysis forms a basis of information credibility analysis. Originality/value - This study formulated the problem of information sender classification. It proposed a method which achieves moderate performance. It also identified some of the issues related to information sender classification.

  380. A Database of Relations between Predicate Argument Structures for Recognizing Textual Entailment and Contradiction 査読有り

    Suguru Matsuyoshi, Koji Murakami, Yuji Matsumoto, Kentaro Inui

    PROCEEDINGS OF THE SECOND INTERNATIONAL SYMPOSIUM ON UNIVERSAL COMMUNICATION 366-373 2008年

    出版者・発行元:IEEE COMPUTER SOC

    DOI: 10.1109/ISUC.2008.31  

    詳細を見る 詳細を閉じる

    In this paper we present a database of logical relations between predicate argument structures (PASs) in Japanese for recognizing relations between statements. We have defined nine logical relations between PASs and manually collected argument structures and logical relations for verbs from definition sentences in a machine-readable Japanese dictionary. In addition, we augmented the relations in our database with a thesaurus of verb argument structures, which identifies synonymy and antonymy between PASs. Our database consists of 29,555 entries and 45,905 relations between PASs. In a preliminary experiment with this database, we constructed a system that recognizes synonymy between PASs in Web documents with a precision of about 0.80.

  381. Experience mining: Building a large-scale database of personal experiences and opinions from web documents 査読有り

    Kentaro Inui, Shuya Abe, Kazuo Hara, Hiraku Morita, Chitose Sao, Megumi Eguchi, Asuka Sumida, Koji Murakami, Suguru Matsuyoshi

    Proceedings - 2008 IEEE/WIC/ACM International Conference on Web Intelligence, WI 2008 314-321 2008年

    DOI: 10.1109/WIIAT.2008.373  

    詳細を見る 詳細を閉じる

    This paper proposes a new UGC-oriented language technology application, which we call experience mining. Experience mining aims at automatically collecting instances of personal experiences as well as opinions from an explosive number of user generated contents (UGCs) such as weblog and forum posts and storing them in an experience database with semantically rich indices. After arguing the technical issues of this new task, we focus on the central problem, factuality analysis, among others and propose a machine learning-based solution as well as the task definition itself. Our empirical evaluation indicates that our factuality analysis task is sufficiently well-defined to achieve a high inter-annotator agreement and our Factorial CRF-based model considerably outperforms the baseline. We also present an application system, which currently stores over 50M experience instances extracted from 150M Japanese blog posts with semantic indices and is scheduled to start serving as an experience search engine for unrestricted users in October. © 2008 IEEE.

  382. Evaluation data and prototype system WISDOM for information credibility analysis 査読有り

    Hisashi Miyamori, Susumu Akamine, Yoshikiyo Kato, Ken Kaneiwa, Kaoru Sumi, Kentaro Inui, Sadao Kurohashi

    INTERNET RESEARCH 18 (2) 155-164 2008年

    出版者・発行元:EMERALD GROUP PUBLISHING LIMITED

    DOI: 10.1108/10662240810862211  

    ISSN:1066-2243

    詳細を見る 詳細を閉じる

    Purpose - The purpose of this paper is to describe evaluation data and a prototype system named WISDOM used for analyzing information credibility based on natural langdage processing. Design/methodology/approach - The authors started the Information Credibility Criteria project in April, 2007, mainly to analyze the credibility of information (text) on the web. The project proposes to capture information credibility based on four criteria (content, sender, appearance, and social valuation) and aims to analyze and organize them logically using natural language processing based on predicate argument structure. Findings - The evaluation data described in this paper were developed as learning and verifying data for these various analysis modules and are composed of manually-annotated data based on each evaluation criteria about several pre-selected topics such as current events and medical issues. The prototype system WISDOM was developed to provide information credibility from different perspectives. Orginality/value - Users will be able to find credible information more reliably by browsing information using different evaluation criteria and conditions provided by the system.

  383. Zero-anaphora resolution by learning rich syntactic pattern features 査読有り

    Ryu Iida, Kentaro Inui, Yuji Matsumoto

    ACM Transactions on Asian Language Information Processing 6 (4) 2007年12月1日

    DOI: 10.1145/1316457.1316458  

    ISSN:1530-0226 1558-3430

    詳細を見る 詳細を閉じる

    We approach the zero-anaphora resolution problem by decomposing it into intrasentential and intersentential zero-anaphora resolution tasks. For the former task, syntactic patterns of zeropronouns and their antecedents are useful clues. Taking Japanese as a target language, we empirically demonstrate that incorporating rich syntactic pattern features in a state-of-the-art learning-based anaphora resolution model dramatically improves the accuracy of intrasentential zero-anaphora, which consequently improves the overall performance of zero-anaphora resolution. © 2007 ACM.

  384. Opinion mining from web documents: extraction and structurization (論文特集:データマイニングと統計数理)

    Kobayashi Nozomi, Inui Kentaro, Matsumoto Yuji

    人工知能学会論文誌 = Transactions of the Japanese Society for Artificial Intelligence : AI 22 227-238 2007年11月1日

    出版者・発行元:The Japanese Society for Artificial Intelligence

    DOI: 10.1527/tjsai.22.227  

    ISSN:1346-0714

    詳細を見る 詳細を閉じる

    The task of opinion extraction and structurization is the key component of opinion mining, which allow Web users to retrieve and summarize people's opinions scattered over the Internet. Our aim is to develop a method for extracting opinions that represent evaluation of concumer products in a structured form. To achieve the goal, we need to consider some issues that are relevant to the extraction task: How the task of opinion extraction and structurization should be designed, and how to extract the opinions which we defined. We define an opinion unit consisting of a quadruple, that is, the opinion holder, the subject being evaluated, the part or the attribute in which it is evaluated, and the evaluation that expresses positive or negative assessment. In this task, we focus on two subtasks (a) extracting subject/aspect-evaluation relations, and (b) extracting subject/aspect-aspect relations, we approach each extraction task using a machine learning-based method. In this paper, we discuss how customer reviews in web documents can be best structured. We also report on the results of our experiments and discuss future directions.

  385. セグメントとリンクに基づくアノテーションツールの設計と実装 査読有り

    野口正樹, 三好健太, 徳永健伸, 飯田龍, 小町守, 乾健太郎

    Vol. (No.) pp. 278-281 2007年3月

  386. 自然言語処理の再挑戦 : 統計的言語処理を超えて(<特集>テキストの可視化と要約) 招待有り

    乾健太郎, 浅原正幸

    知能と情報 : 日本知能情報ファジィ学会誌 : journal of Japan Society for Fuzzy Theory and Intelligent Informatics 18 (5) 669-681 2006年10月

    ISSN:1347-7986

  387. グルメ情報を含む Web 文書からのユーザ指向型評判情報抽出システムの開発

    新井 イスマイル, 飯田 龍, 小林 のぞみ, 乾 健太郎, 藤川 和利, 砂原 秀樹

    情報処理学会, マルチメディア, 分散, 協調とモバイル(DICOMO2006)シンポジウム論文集 953-956 2006年7月

  388. クラス指向事例収集手法による言い換えコーパスの構築 査読有り

    藤田 篤, 乾 健太郎

    自然言語処理 = Journal of natural language processing 13 (3) 133-150 2006年7月

    出版者・発行元:言語処理学会

    DOI: 10.5715/jnlp.13.3_133  

    ISSN:1340-7619 2185-8314

  389. 動詞項構造辞書への大規模用例付与 査読有り

    平野 徹, 飯田 龍, 藤田 篤, 乾 健太郎, 松本 裕治

    自然言語処理 = Journal of natural language processing 13 (3) 113-132 2006年7月

    出版者・発行元:言語処理学会

    DOI: 10.5715/jnlp.13.3_113  

    ISSN:1340-7619 2185-8314

  390. 語彙概念構造に基づく言い換え生成 : 機能動詞構文の言い換えを例題に(自然言語) 査読有り

    藤田篤, 降幡建太郎, 乾健太郎, 松本裕治

    情報処理学会論文誌 47 (6) 1963-1975 2006年6月

    出版者・発行元:一般社団法人情報処理学会

    ISSN:1882-7764

    詳細を見る 詳細を閉じる

    本論文では,語彙概念構造(LCS)に基づく言い換え生成モデルを提案する.LCSは語彙意味論の体系の一つであり,動詞の様々な統語・意味的な特性に対応する意味述語を組み合わせて動詞の意味を記述する枠組みである.提案するモデルは,LCSが表す意味構造の変形操作および語彙意味論的制約に基づいて言い換えを生成する.本論文では,例題として日本語の機能動詞構文の言い換えを取り上げ,言い換えの生成実験を通じてモデルおよび現存のLCS辞書の性能を検証する.

  391. Exploiting Syntactic Patterns as Clues in Zero-Anaphora Resolution 査読有り

    Ryu Iida, Kentaro Inui, Yuji Matsumoto

    COLING/ACL 2006, VOLS 1 AND 2, PROCEEDINGS OF THE CONFERENCE 625-632 2006年

    出版者・発行元:ASSOC COMPUTATIONAL LINGUISTICS-ACL

    詳細を見る 詳細を閉じる

    We approach the zero-anaphora resolution problem by decomposing it into intra-sentential and inter-sentential zero-anaphora resolution. For the former problem, syntactic patterns of the appearance of zero-pronouns and their antecedents are useful clues. Taking Japanese as a target language, we empirically demonstrate that incorporating rich syntactic pattern features in a state-of-the-art learning-based anaphora resolution model dramatically improves the accuracy of intra-sentential zero-anaphora, which consequently improves the overall performance of zero-anaphora resolution.

  392. 意見抽出のための評価表現の収集 査読有り

    小林 のぞみ, 乾 健太郎, 松本 裕治, 立石 健二, 福島 俊一

    自然言語処理 = Journal of natural language processing 12 (3) 203-222 2005年7月

    出版者・発行元:言語処理学会

    DOI: 10.5715/jnlp.12.3_203  

    ISSN:1340-7619 2185-8314

  393. 英国で見つけた歴史の試験問題に思う 招待有り

    乾 健太郎

    人工知能学会誌 = Journal of Japanese Society for Artificial Intelligence 20 (2) 243-244 2005年3月

    ISSN:0912-8085

  394. 最尤先行詞候補を用いた日本語名詞句同一指示解析(自然言語)(<特集>情報システム論文) 査読有り

    飯田龍, 乾健太郎, 松本裕治, 関根聡

    情報処理学会論文誌 46 (3) 831-844 2005年3月

    出版者・発行元:一般社団法人情報処理学会

    ISSN:1882-7764

    詳細を見る 詳細を閉じる

    日本語における名詞句の同一指示関係を同定するための新しい解析手法を提案する.従来の同一指示解析方法は定名詞か否かを判定する際に局所文脈の情報しか参照していないという問題や, 非照応詞と先行詞候補との関係を訓練事例として抽出していないため, 非照応詞が棄却されることが保証されないという問題が存在する.また, 日本語の場合は, 英語などの言語と比較して冠詞の情報がないため同一指示関係の解析の問題が難しくなると考えられる.そこで, 本稿では, 照応詞の候補に対して先行詞となる可能性のある候補を提示することで, より広い文脈の情報を参照して照応詞か否かの分類問題を解くための手法を提案する.この手法では, 非照応詞に対しても最も先行詞らしい候補を見せることで極端に負例が多くならないという利点もある.この提案手法を用いて日本語名詞句同一指示関係の同定実験を行い, 先行研究の機械学習を用いた手法より精度良く同一指示関係の同定ができたことを報告する.

  395. 最尤先行詞候補を用いた日本語名詞句同一指示解析 査読有り

    飯田 龍, 乾 健太郎, 松本 裕治, 関根 聡

    情報処理学会論文誌 = Transactions of Information Processing Society of Japan 46 (3) 831-844 2005年3月

  396. Acquiring causal knowledge from text using the connective marker tame 査読有り

    Takashi Inui, Kentaro Inui, Yuji Matsumoto

    ACM Transactions on Asian Language Information Processing 4 (4) 435-474 2005年

    DOI: 10.1145/1113308.1113313  

    ISSN:1530-0226

    詳細を見る 詳細を閉じる

    In this paper, we deal with automatic knowledge acquisition from text, specifically the acquisition of causal relations. A causal relation is the relation existing between two events such that one event causes (or enables) the other event, such as "hard rain causes flooding" or "taking a train requires buying a ticket." In previous work these relations have been classified into several types based on a variety of points of view. In this work, we consider four types of causal relations-cause, effect, precond(ition) and means-mainly based on agents' volitionality, as proposed in the research field of discourse understanding. The idea behind knowledge acquisition is to use resultative connective markers, such as "because," "but," and "if" as linguistic cues. However, there is no guarantee that a given connective marker always signals the same type of causal relation. Therefore, we need to create a computational model that is able to classify samples according to the causal relation. To examine how accurately we can automatically acquire causal knowledge, we attempted an experiment using Japanese newspaper articles, focusing on the resultative connective "tame." By using machine-learning techniques, we achieved 80% recall with over 95% precision for the cause, precond, and means relations, and 30% recall with 90% precision for the effect relation. Furthermore, the classification results suggest that one can expect to acquire over 27,000 instances of causal relations from 1 year of Japanese newspaper articles. © 2005 ACM.

  397. On the issue of combining anaphoricity determination and antecedent identification in anaphora resolution 査読有り

    Ryu Iida, Kentaro Inui, Yuji Matsumoto

    Proceedings of 2005 IEEE International Conference on Natural Language Processing and Knowledge Engineering, IEEE NLP-KE'05 2005 244-249 2005年

    DOI: 10.1109/NLPKE.2005.1598742  

    詳細を見る 詳細を閉じる

    We propose a machine learning-based approach to noun phrase anaphora resolution that combines the advantages of previous learning-based models while overcoming their drawbacks. Our anaphora resolution process reverses the order of the steps in the classification-and-search model proposed by Ng and Cardie [12], but inherits all the advantages of that model. We conducted experiments on resolving noun phrase anaphora in Japanese. The results show that with the classification-and-search based modifications, our proposed model outperforms earlier learning-based approaches. © 2005 IEEE.

  398. チャット対話における発言間の継続関係と応答関係の同定 査読有り

    徳永 泰浩, 乾 健太郎, 松本 裕治

    自然言語処理 = Journal of natural language processing 12 (1) 79-105 2005年1月

    出版者・発行元:言語処理学会

    DOI: 10.5715/jnlp.12.79  

    ISSN:1340-7619 2185-8314

  399. 言い換え技術に関する研究動向 招待有り

    乾 健太郎, 藤田 篤

    自然言語処理 = Journal of natural language processing 11 (5) 151-198 2004年10月

    出版者・発行元:言語処理学会

    DOI: 10.5715/jnlp.11.5_151  

    ISSN:1340-7619 2185-8314

  400. 自動生成された言い換え文における不適格な動詞格構造の検出(自然言語) 査読有り

    藤田篤, 乾健太郎, 松本裕治

    情報処理学会論文誌 45 (4) 1176-1187 2004年4月

    出版者・発行元:一般社団法人情報処理学会

    ISSN:1882-7764

    詳細を見る 詳細を閉じる

    本論文では,語彙・構文的言い換えにおいて頻繁に生じる動詞格構造の不整合を自動的に検出する方法を提案する.我々は,コーパスから獲得した大規模な正例に基づいて格構造の適格さを定量化する確率的言語モデルと,人手で収集した小規模な負例に基づいて格構造の不適格さを定量化するモデルを構築し,これら2つを混合し,正例のみに基づく言語モデルと比較して精度の高い誤り検出器を実現した.また,誤り検出に対して貢献度が高い負例を効率良く収集するために能動学習を試行した.

  401. 文脈的手がかりを考慮した機械学習による日本語ゼロ代名詞の先行詞同定(自然言語) 査読有り

    飯田龍, 乾健太郎

    情報処理学会論文誌 45 (3) 906-918 2004年3月

    出版者・発行元:一般社団法人情報処理学会

    ISSN:1882-7764

    詳細を見る 詳細を閉じる

    センタリング理論のような言語学的な知見を,機械学習を用いた照応解析に統合する一方法を提案する.従来の照応解析手法は,大きく規則ベースの手法と統計的な手法に分類でき,それぞれ独立に研究がなされてきた.規則ベースの手法では,言語学的知見に基づき人手で規則を記述するが,照応現象を包括的にとらえる規則を書き尽くすことは困難である.一方,機械学習に基づく手法では,人手では扱うことのできない規則の組合せを考慮できるが,言語学的知見を有効に活用していない.これら2つの手法をうまく統合することによって,両者の利点を同時に引き出すことができれば,精度の向上がさらに期待できる.本論文では2つの手法の統合を目指し,具体的な方法として(i)センタリング理論に基づく局所的な文脈を考慮した素性(センタリング素性)の導入,および(ii)先行詞候補間を比較するモデル(トーナメントモデル)の2点を提案する.この提案手法を用いて日本語ゼロ代名詞の同定を行い,先行研究の機械学習を用いた手法より精度良く先行詞の同定かできたことを報告する.

  402. 接続標識「ため」に基づく文書集合からの因果関係知識の自動獲得(自然言語) 査読有り

    乾孝司, 乾健太郎, 松本裕治

    情報処理学会論文誌 45 (3) 919-933 2004年3月

    出版者・発行元:一般社団法人情報処理学会

    ISSN:1882-7764

    詳細を見る 詳細を閉じる

    人間のような深い言語理解能力を工学的に実現することを妨げている1つの原因として,計算機で利用可能な大量の常識的知識をいかに構築するかという問題がある.本論文では,常識的知識のうち,因果関係に関する知識に着目し,大規模な電子化文書集合から因果関係知識を自動的に獲得する方法について論じる.論文では,まず,因果関係に立つ2つの出来事がそれぞれ行為であるか事態であるかに基づき,cause, effect, precond (ition), meansの4つの因果関係を定義した.接続標識「ため」を合む複文からこれら4つの因果関係への自動分類実験を行い,cause, precond, meansの各関係について,80%の再現率で95%以上の分類精度を達成した.また,effect関係については,30%の再現率で90%の分類精度を達成した.さらに,新聞記事1年分の文書集合に本手法を適用することにより,およそ27,000件を超える因果知識が獲得できる見積りを得た.

  403. 語彙・構文的言い換えにおける変換誤りの分析(自然言語)(<特集>「インタラクション:理論,技術,応用,評価」) 査読有り

    藤田篤, 乾健太郎

    情報処理学会論文誌 44 (11) 2826-2838 2003年11月

    詳細を見る 詳細を閉じる

    本論文では,世界知識や発話の文脈に関係なく言語学的知識で実現可能な語彙・構文的言い換えを対象とし,これらを構文トランスファ方式で実現する際に生じる変換誤りについて述べる.本研究では,次の2段階の調査を行った.(1)まず,先行研究で示されている知識・知見に基づいて構文トランスファによって生じる変換誤りを洗い出し,その種類を整理・分類し,(2)次に,さまざまな種類の構文トランスファ規則を用いて言い換え事例を生成し,事例分析に基づいて,各種変換誤りが生じる傾向を調査した.これらの調査によって,多くの変換誤りが,構文トランスファ規則の種類に依存することなく生じることが明らかになった.また,問題の性質を分析し,解消に向けての課題を整理した結果,まずは頻繁に生じる傾向がある形態的な不適格性,および構文的な不適格性を優先的に解消すべきであるという結論を得た.

  404. What kinds and amounts of causal knowledge can be acquired from text by using connective markers as clues? 査読有り

    T Inui, K Inui, Y Matsumoto

    DISCOVERY SCIENCE, PROCEEDINGS 2843 180-193 2003年

    出版者・発行元:SPRINGER-VERLAG BERLIN

    ISSN:0302-9743

    詳細を見る 詳細を閉じる

    This paper reports the results of our ongoing research into the automatic acquisition of causal knowledge. We created a new typology for expressing the causal relations - cause, effect, precond(ition) and means - based mainly on the volitionality of the related events. From our experiments using the Japanese resultative connective "tame", we achieved 80% recall with over 95% precision for the cause, precond and means relations, and 30% recall with 90% precision for the effect relation. The results indicate that over 27,000 instances of causal relations can be acquired from one year of Japanese newspaper articles.

  405. 複数のパーザを利用した統計的部分係り受け解析 査読有り

    乾孝司, 乾健太郎

    情報処理学会論文誌 42 (12) 3160-3172 2001年12月

    出版者・発行元:一般社団法人情報処理学会

    ISSN:1882-7764

    詳細を見る 詳細を閉じる

    我々はこれまでに, 信頼のおける部分だけを出力し被覆率を犠牲とする代償として正解率を向上させる統計的部分解析手法の調査を進めてきた.本稿では, さらにこの考えに委員会方式という概念を統合した枠組みを提案し, その評価を行った.委員会方式とは, 複数の解析器(委員)の出力解を組み合わせることにより解析精度の向上をはかる手法である.ここでは, 各委員から得られる解析結果に基づき委員会で多数決により統計的部分解析を行うために, 従来の基本的な委員会方式に3つの拡張を施した:(1)解析器(委員)が推定した係り受け確率を票の重みと見なして重みつきの票を投じる確率的投票, (2)委員間での票の重みの信頼性を標準化する重み標準化, (3)各係り文節に対して2位以下の係り先候補にも重みつきの票を投じる多重投票.既存の5つの統計的解析器を用いて, 京大コーパスを対象データとする解析実験を行った.その結果, 委員の組合せによって精度変化には多少の揺れがあるものの, 総合的には提案した枠組みおよび3つの拡張が解析精度の向上に有効に作用する見通しを得た.

  406. Selective sampling for example-based word sense disambiguation 査読有り

    A Fujii, K Inui, T Tokunaga, H Tanaka

    COMPUTATIONAL LINGUISTICS 24 (4) 573-597 1998年12月

    出版者・発行元:M I T PRESS

    ISSN:0891-2017

    詳細を見る 詳細を閉じる

    This paper proposes an efficient example sampling method for example-based word sense disambiguation systems. To construct a database of practical size, a considerable overhead for manual sense disambiguation (overhead for supervision) is required. In addition, the time complexity gf searching a large-sized database poses a considerable problem (overhead for search). To counter these problems, our method selectively samples a smaller-sized effective subset from a given example set for use in word sense disambiguation. Our method is characterized by the reliance on the notion of training utility: the degree to which each example is informative for future example sampling when used for the training of the system. The system progressively collects examples by selecting those with greatest utility. The paper reports the effectiveness of our method through experiments on about one thousand sentences. Compared to experiments with other example sampling methods, our method reduced both the overhead for supervision and the overhead for search, without the degeneration of the performance of the system.

  407. Towards the Application of Word Sense Disambiguation to Information Extraction 査読有り

    Atsushi Fujii, Kentaro Inui, Timothy Baldwin, Takenobu Tokunaga, Hozumi Tanaka

    Proceedings of the International Workshop on Lexically Driven Information Extraction Vol. (No.) pp. 39-49 1997年7月

  408. Case Contribution in Example-Based Verb Sense Disambiguation 査読有り

    Atsushi Fujii, Kentaro Inui, Takenobu Tokunaga, Hozumi Tanaka

    自然言語処理 Vol. 4 (No. 2) pp. 111-123 1997年4月

  409. Selective Sampling of Effective Example Sentence Sets for Word Sense Disambiguation 査読有り

    Atsushi Fujii, Kentaro Inui, Takenobu Tokunaga, Hozumi Tanaka

    Proceedings of the 4th Workshop on Very Large Corpora Vol. (No.) pp. 56-69 1996年8月

  410. To what extent does case contribute to verb sense disambiguation? 査読有り

    Atsushi Fujii, Kentaro Inui, Takenobu Tokunaga, Hozumi Tanaka

    Proceedings of the 16th International Conference on Computational Linguistics Vol. (No.) pp. 59-64 1996年8月

  411. Dependency-directed control of text generation using functional unification grammar 査読有り

    K Inui, T Tokunaga, H Tanaka

    NEW GENERATION COMPUTING 14 (2) 169-193 1996年

    出版者・発行元:SPRINGER VERLAG

    DOI: 10.1007/BF03037498  

    ISSN:0288-3635

    詳細を見る 詳細を閉じる

    In text generation, various kinds of choices need to be decided. In conventional frameworks, which we call one-path generation frameworks, choices are made in an order carefully designed in advance. In general, however, since choices depend on one another, it is difficult to make optimal decisions in such frameworks. Our approach to this issue is to introduce the revision process into the overall generation process. In our framework, revision of output texts is realized as dependency-directed backtracking (DDB). As well as Justification-based Truth Maintenance System (JTMS), we maintain dependencies among choices in a dependency network. In this paper,we propose an efficient implementation of DDB for text generation using functional unification grammar (FUG). We use bindings of logical variables in Prolog and destructive argument substitutions to decrease the overhead of handling a dependency network. This paper describes the algorithm in detail and shows the results of preliminary experiments to demonstrate the performance of our implementation.

  412. Text revision: A model and its implementation 査読有り

    Kentaro Inui, Takenobu Tokunaga, Hozumi Tanaka

    Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) 587 215-230 1992年

    出版者・発行元:Springer Verlag

    DOI: 10.1007/3-540-55399-1_15  

    ISSN:1611-3349 0302-9743

    詳細を見る 詳細を閉じる

    To generate good text, many kinds of decisions should be made. Many researchers have spent much time searching for the architecture that would determine a proper order for these decisions. However, even if such an architecture is found, there are still certain kinds of problems that are difficult to consider during the generation process. Those problems can be more easily detected and solved by introducing a revision process after generation. In this paper, we argue the importance of text revision with respect to natural language generation, and propose a computational model of text revision. We also discuss its implementation issues and describe an experimental Japanese text generation system, WeiveR.

  413. Text Simplification for Reading Assistance: A Project Note 査読有り

    Kentaro Inui, Atsushi Fujita, Tetsuro Takahashi, Ryu Iida, Tomoya Iwakura

    詳細を見る 詳細を閉じる

    This paper describes our ongoing research project on text simplification for congenitally deaf people. Text simplification we are aiming at is the task of offering a deaf reader a syntactic and lexical paraphrase of a given text for assisting her/him to understand what it means.

  414. Annotating a Japanese Text Corpus with Predicate-Argument and Coreference Relations 査読有り

    Ryu Iida, Mamoru Komachi, Kentaro Inui, Yuji Matsumoto

    詳細を見る 詳細を閉じる

    In this paper, we discuss how to annotate coreference and predicate-argument relations in Japanese written text. There have been research activities for building Japanese text corpora annotated with coreference and predicate-argument relations as are done in the Kyoto Text Corpus version 4.0 (Kawahara et al., 2002) and the GDA-Tagged Corpus (Hasida, 2005). However, there is still much room for refining their specifications. For this reason, we discuss issues in annotating these two types of relations, and propose a new specification for each. In accordance with the specification, we built a large-scaled annotated corpus, and examined its reliability. As a result of our current work, we have released an annotated corpus named the NAIST Text Corpus1, which is used as the evaluation data set in the coreference and zero-anaphora resolution tasks in Iida et al. (2005) and Iida et al. (2006). 1

  415. c○2009 The Association for Computational Linguistics and The Asian Federation of Natural Language Processing 査読有り

    Of The Afnlp, Toh Tuck Link, Susumu Akamine, Daisuke Kawahara, Yoshikiyo Kato, Tetsuji Nakagawa, Kentaro Inui

    詳細を見る 詳細を閉じる

    ii Preface Welcome to the proceedings of the demo session. This volume contains the abstracts of the software demonstrations presented at the combined 47th Annual Meeting of the Association for Computational Linguistics and 4th International Joint Conference on Natural Language Processing, held in Suntec, Singapore, on August 3. The demonstrations program offers the presentation of early research prototypes as well as interesting mature systems. The demo chairs and the members of the program committee received 23 submissions, 11 of which were selected for inclusion in the program after review by at least two members of the program committee and on-line discussions for nine boundary papers. We would like to thank the members of our program committee for their excellent job in reviewing the submissions and providing their support in the decisive discussions. Co-chairs:

  416. Dependency-directed unification of Functional Unification Grammar in text generation 査読有り

    Kentaro Inui, Takenobu Tokunaga, Tanaka Hozumi

    詳細を見る 詳細を閉じる

    In text generation, various kinds of choices need to be decided. In the conventional framework, which can be called &quot;one-path generation framework,&quot; these choices are decided in an order designed carefully in advance. However, many researchers have pointed out that the choices, generally, depend on one another and the one-path generation framework cannot handle these interdependencies sufficiently. Our previous paper proposed introducing a revision process into text generation for solving this problem. In our framework, the overall generation process consists of the initial generation process, followed by the revision process. The revision process gives us opportunities to change choices that have already been made. In general, a change in a choice point may cause changes in other choice points, and such dependencies can be managed by Truth Maintenance System (TMS). However, it is well known that dependency network management in TMS requires some computational overhead in general. We n...

︎全件表示 ︎最初の5件までを表示

MISC 252

  1. ニューラル記号推論における推論過程の教示方法

    青木洋一, 工藤慧音, BRASSARD Ana, BRASSARD Ana, 栗林樹生, 栗林樹生, 吉川将司, 坂口慶祐, 坂口慶祐, 乾健太郎, 乾健太郎

    言語処理学会年次大会発表論文集(Web) 29th 2023年

    ISSN:2188-4420

  2. 算術問題におけるニューラルモデルの構成的推論能力

    工藤慧音, 青木洋一, 栗林樹生, 栗林樹生, BRASSARD Ana, BRASSARD Ana, 吉川将司, 坂口慶祐, 坂口慶祐, 乾健太郎, 乾健太郎

    言語処理学会年次大会発表論文集(Web) 29th 2023年

    ISSN:2188-4420

  3. 文章構造グラフを用いた国語記述式答案への自動フィードバック生成

    岩瀬裕哉, 舟山弘晃, 松林優一郎, 乾健太郎, 岩瀬裕哉, 舟山弘晃, 松林優一郎, 乾健太郎

    言語処理学会年次大会発表論文集(Web) 29th 2023年

    ISSN:2188-4420

  4. 日本語話者の項省略判断に関するアノテーションとモデリング

    石月由紀子, 石月由紀子, 栗林樹生, 栗林樹生, 松林優一郎, 松林優一郎, 笹野遼平, 笹野遼平, 乾健太郎, 乾健太郎

    言語処理学会年次大会発表論文集(Web) 29th 2023年

    ISSN:2188-4420

  5. 記述式答案採点モデルの採点基準に対する整合性の検証

    浅妻佑弥, 浅妻佑弥, 舟山弘晃, 舟山弘晃, 松林優一郎, 松林優一郎, 水本智也, 乾健太郎, 乾健太郎

    言語処理学会年次大会発表論文集(Web) 29th 2023年

    ISSN:2188-4420

  6. シフト付き絶対位置埋め込み

    清野舜, 清野舜, 小林颯介, 小林颯介, 鈴木潤, 鈴木潤, 乾健太郎, 乾健太郎

    言語処理学会年次大会発表論文集(Web) 28th 2022年

    ISSN:2188-4420

  7. 記述式答案自動採点における確信度推定とその役割

    舟山弘晃, 舟山弘晃, 佐藤汰亮, 佐藤汰亮, 松林優一郎, 松林優一郎, 水本智也, 水本智也, 鈴木潤, 鈴木潤, 乾健太郎, 乾健太郎

    言語処理学会年次大会発表論文集(Web) 28th 2022年

    ISSN:2188-4420

  8. 規範的な日本語日常対話コーパスの設計

    赤間怜奈, 赤間怜奈, 磯部順子, 磯部順子, 鈴木潤, 鈴木潤, 乾健太郎, 乾健太郎

    言語処理学会年次大会発表論文集(Web) 28th 2022年

    ISSN:2188-4420

  9. 論述リビジョンのためのメタ評価基盤

    三田雅人, 坂口慶祐, 萩原正人, 萩原正人, 水本智也, 水本智也, 鈴木潤, 鈴木潤, 乾健太郎, 乾健太郎

    言語処理学会年次大会発表論文集(Web) 28th 2022年

    ISSN:2188-4420

  10. 機械翻訳モデルの頑健性評価に向けた言語現象毎データセットの構築と分析

    藤井諒, 三田雅人, 三田雅人, 阿部香央莉, 塙一晃, 塙一晃, 森下睦, 森下睦, 鈴木潤, 鈴木潤, 乾健太郎, 乾健太郎

    自然言語処理(Web) 28 (2) 2021年

    ISSN:2185-8314

  11. データ指向型アプローチによるクローズドブック質問応答

    鈴木正敏, 鈴木正敏, 松田耕史, 松田耕史, 大内啓樹, 大内啓樹, 鈴木潤, 鈴木潤, 乾健太郎, 乾健太郎

    自然言語処理(Web) 28 (1) 2021年

    ISSN:2185-8314

  12. 対話システムの先読み能力実現に向けた未来の展開まで生成する学習戦略の提案と分析

    岸波洋介, 赤間怜奈, 赤間怜奈, 佐藤志貴, 鈴木潤, 鈴木潤, 徳久良子, 乾健太郎, 乾健太郎

    人工知能学会全国大会(Web) 35th 2021年

  13. 目標指向対話のためのサブゴール設計に基づく発話系列生成

    岸波洋介, 赤間怜奈, 赤間怜奈, 佐藤志貴, 徳久良子, 鈴木潤, 鈴木潤, 乾健太郎, 乾健太郎

    人工知能学会言語・音声理解と対話処理研究会資料 93rd 2021年

    ISSN:0918-5682

  14. 予測の正確な言語モデルがヒトらしいとは限らない

    栗林樹生, 栗林樹生, 大関洋平, 大関洋平, 伊藤拓海, 伊藤拓海, 吉田遼, 浅原正幸, 乾健太郎, 乾健太郎

    言語処理学会年次大会発表論文集(Web) 27th 2021年

    ISSN:2188-4420

  15. 日本語の読みやすさに対する情報量に基づいた統一的な解釈

    栗林樹生, 栗林樹生, 大関洋平, 大関洋平, 伊藤拓海, 伊藤拓海, 吉田遼, 浅原正幸, 乾健太郎, 乾健太郎

    言語処理学会年次大会発表論文集(Web) 27th 2021年

    ISSN:2188-4420

  16. Transformerの文脈を混ぜる作用と混ぜない作用

    小林悟郎, 栗林樹生, 栗林樹生, 横井祥, 横井祥, 乾健太郎, 乾健太郎

    言語処理学会年次大会発表論文集(Web) 27th 2021年

    ISSN:2188-4420

  17. 人と言語モデルが捉える文の主題

    藤原吏生, 栗林樹生, 栗林樹生, 乾健太郎, 乾健太郎

    言語処理学会年次大会発表論文集(Web) 27th 2021年

    ISSN:2188-4420

  18. 物語におけるイベントの顕現性推定と物語類似性計算への応用

    大竹孝樹, 横井祥, 横井祥, 井之上直也, 井之上直也, 高橋諒, 高橋諒, 栗林樹生, 栗林樹生, 乾健太郎, 乾健太郎

    言語処理学会年次大会発表論文集(Web) 27th 2021年

    ISSN:2188-4420

  19. 英語学習者のための解説文生成手法の調査

    塙一晃, 塙一晃, 永田亮, 永田亮, 永田亮, 乾健太郎, 乾健太郎

    言語処理学会年次大会発表論文集(Web) 27th 2021年

    ISSN:2188-4420

  20. オンラインコミュニティにおける単語頻度の通時的変化を利用した新語リストの獲得

    阿部香央莉, 松田耕史, 松田耕史, 吉川将司, 吉川将司, 乾健太郎, 乾健太郎

    言語処理学会年次大会発表論文集(Web) 27th 2021年

    ISSN:2188-4420

  21. ニューラルネットが学習する意味表現は体系性を持つか

    谷中瞳, 峯島宏次, 乾健太郎, 乾健太郎

    言語処理学会年次大会発表論文集(Web) 27th 2021年

    ISSN:2188-4420

  22. 単語埋め込みの決定的縮約

    仲村祐希, 鈴木潤, 鈴木潤, 高橋諒, 高橋諒, 乾健太郎, 乾健太郎

    言語処理学会年次大会発表論文集(Web) 27th 2021年

    ISSN:2188-4420

  23. 単一事例エキスパートの統合によるドメイン適応

    清野舜, 清野舜, 小林颯介, 小林颯介, 鈴木潤, 鈴木潤, 乾健太郎, 乾健太郎

    言語処理学会年次大会発表論文集(Web) 27th 2021年

    ISSN:2188-4420

  24. オープンドメイン質問応答における解答可能性判別の役割

    鈴木正敏, 鈴木正敏, 松田耕史, 松田耕史, 大内啓樹, 大内啓樹, 鈴木潤, 鈴木潤, 乾健太郎, 乾健太郎

    言語処理学会年次大会発表論文集(Web) 27th 2021年

    ISSN:2188-4420

  25. 事例ベース依存構造解析のための依存関係表現学習

    大内啓樹, 鈴木潤, 鈴木潤, 小林颯介, 小林颯介, 横井祥, 横井祥, 栗林樹生, 栗林樹生, 吉川将司, 吉川将司, 乾健太郎, 乾健太郎

    言語処理学会年次大会発表論文集(Web) 27th 2021年

    ISSN:2188-4420

  26. 対話システムの先読み能力を分析可能なタスクの検討

    岸波洋介, 赤間怜奈, 赤間怜奈, 佐藤志貴, 鈴木潤, 鈴木潤, 乾健太郎, 乾健太郎

    言語処理学会年次大会発表論文集(Web) 27th 2021年

    ISSN:2188-4420

  27. 対話システムの矛盾応答の生成に対する脆弱性の分析

    佐藤志貴, 赤間怜奈, 赤間怜奈, 大内啓樹, 鈴木潤, 鈴木潤, 乾健太郎, 乾健太郎

    言語処理学会年次大会発表論文集(Web) 27th 2021年

    ISSN:2188-4420

  28. Langsmith:人とシステムの協働による論文執筆

    伊藤拓海, 伊藤拓海, 栗林樹生, 栗林樹生, 日高雅俊, 鈴木潤, 鈴木潤, 乾健太郎, 乾健太郎

    言語処理学会年次大会発表論文集(Web) 27th 2021年

    ISSN:2188-4420

  29. 項目採点技術に基づいた和文英訳答案の自動採点

    菊地正弥, 菊地正弥, 尾中大介, 尾中大介, 舟山弘晃, 舟山弘晃, 松林優一郎, 松林優一郎, 乾健太郎, 乾健太郎

    言語処理学会年次大会発表論文集(Web) 27th 2021年

    ISSN:2188-4420

  30. 事前学習とfinetuningの類似性に基づくゼロ照応解析

    今野颯人, 清野舜, 清野舜, 松林優一郎, 松林優一郎, 大内啓樹, 乾健太郎, 乾健太郎

    言語処理学会年次大会発表論文集(Web) 27th 2021年

    ISSN:2188-4420

  31. 句の呼応と話題の一貫性に着目した低品質対話データの教師なしフィルタリング

    赤間怜奈, 鈴木潤, 横井祥, 乾健太郎, 赤間怜奈, 鈴木潤, 横井祥, 乾健太郎

    言語処理学会年次大会発表論文集(Web) 26th 2020年

    ISSN:2188-4420

  32. 言語モデルによる物語中のイベントの顕現性推定

    大竹孝樹, 横井祥, 横井祥, 井之上直也, 井之上直也, 高橋諒, 高橋諒, 栗林樹生, 栗林樹生, 乾健太郎, 乾健太郎

    言語処理学会年次大会発表論文集(Web) 26th 2020年

    ISSN:2188-4420

  33. 高信頼度な文法誤り解説生成のための生成制御手法

    塙一晃, 塙一晃, 永田亮, 永田亮, 乾健太郎, 乾健太郎

    人工知能学会全国大会(Web) 34th 2020年

  34. 日本語語順分析に言語モデルを用いることの妥当性について

    栗林樹生, 栗林樹生, 伊藤拓海, 伊藤拓海, 鈴木潤, 鈴木潤, 乾健太郎, 乾健太郎

    言語処理学会年次大会発表論文集(Web) 26th 2020年

    ISSN:2188-4420

  35. テキストを通して世界を見る:機械読解における常識的推論のための画像説明文の評価

    GALVAN-SOSA Diana, 西田京介, 松田耕史, 松田耕史, 鈴木潤, 鈴木潤, 乾健太郎, 乾健太郎

    言語処理学会年次大会発表論文集(Web) 26th 2020年

    ISSN:2188-4420

  36. 文法誤り訂正のための自己改良戦略に基づくノイズ除去

    三田雅人, 三田雅人, 清野舜, 清野舜, 金子正弘, 金子正弘, 鈴木潤, 鈴木潤, 乾健太郎, 乾健太郎

    言語処理学会年次大会発表論文集(Web) 26th 2020年

    ISSN:2188-4420

  37. スパン間の類似性に基づく事例ベース構造予測

    大内啓樹, 大内啓樹, 鈴木潤, 鈴木潤, 小林颯介, 小林颯介, 横井祥, 横井祥, 栗林樹生, 栗林樹生, 乾健太郎, 乾健太郎

    言語処理学会年次大会発表論文集(Web) 26th 2020年

    ISSN:2188-4420

  38. 評価データのクラスタリングを用いた記述式答案自動採点のためのトランズダクティブ学習

    佐藤俊, 佐々木翔大, 佐々木翔大, 大内啓樹, 大内啓樹, 鈴木潤, 鈴木潤, 乾健太郎, 乾健太郎

    言語処理学会年次大会発表論文集(Web) 26th 2020年

    ISSN:2188-4420

  39. 大規模疑似データを用いた高性能文法誤り訂正モデルの構築

    清野舜, 清野舜, 鈴木潤, 鈴木潤, 三田雅人, 三田雅人, 水本智也, 水本智也, 乾健太郎, 乾健太郎

    言語処理学会年次大会発表論文集(Web) 26th 2020年

    ISSN:2188-4420

  40. 単一評価サンプルのためのトランズダクティブ学習

    佐々木翔大, 佐々木翔大, 大内啓樹, 大内啓樹, 鈴木潤, 鈴木潤, BRASSARD Ana, BRASSARD Ana, 乾健太郎, 乾健太郎

    言語処理学会年次大会発表論文集(Web) 26th 2020年

    ISSN:2188-4420

  41. 超球面上での最適輸送に基づく文類似性尺度

    横井祥, 高橋諒, 赤間怜奈, 鈴木潤, 乾健太郎, 横井祥, 高橋諒, 赤間怜奈, 鈴木潤, 乾健太郎

    言語処理学会年次大会発表論文集(Web) 26th 2020年

    ISSN:2188-4420

  42. ベクトル長に基づく自己注意機構の解析

    小林悟郎, 栗林樹生, 栗林樹生, 横井祥, 横井祥, 鈴木潤, 鈴木潤, 乾健太郎, 乾健太郎

    言語処理学会年次大会発表論文集(Web) 26th 2020年

    ISSN:2188-4420

  43. 対話応答選択による対話応答生成モデルの評価

    佐藤志貴, 赤間怜奈, 赤間怜奈, 大内啓樹, 大内啓樹, 鈴木潤, 鈴木潤, 乾健太郎, 乾健太郎

    言語処理学会年次大会発表論文集(Web) 26th 2020年

    ISSN:2188-4420

  44. ニューラル対話応答生成のための言語非依存な低品質対話データフィルタリング法の提案と分析

    赤間怜奈, 赤間怜奈, 横井祥, 横井祥, 鈴木潤, 鈴木潤, 乾健太郎, 乾健太郎

    人工知能学会全国大会(Web) 34th 2020年

  45. ユーザ生成コンテンツの高品質な自動翻訳に向けた言語現象の体系的分析

    藤井諒, 三田雅人, 三田雅人, 阿部香央莉, 阿部香央莉, 塙一晃, 塙一晃, 森下睦, 鈴木潤, 鈴木潤, 乾健太郎, 乾健太郎

    人工知能学会全国大会(Web) 34th 2020年

  46. 単語埋め込みのノルムと方向ベクトルを区別した文間最適輸送コスト

    横井祥, 横井祥, 高橋諒, 高橋諒, 赤間怜奈, 赤間怜奈, 鈴木潤, 鈴木潤, 乾健太郎, 乾健太郎

    人工知能学会全国大会(Web) 34th 2020年

  47. ILYS aoba bot:大規模ニューラル応答生成モデルとルールベースを統合した雑談対話システム

    藤原吏生, 岸波洋介, 今野颯人, 佐藤志貴, 佐藤汰亮, 宮脇峻平, 加藤拓真, 鈴木潤, 鈴木潤, 乾健太郎, 乾健太郎

    人工知能学会言語・音声理解と対話処理研究会資料 90th 2020年

    ISSN:0918-5682

  48. 反復改良法を用いた日本語述語項構造解析

    宮脇峻平, 清野舜, 清野舜, 松林優一郎, 松林優一郎, 今野颯人, 高橋諒, 高橋諒, 大内啓樹, 大内啓樹, 乾健太郎, 乾健太郎

    言語処理学会年次大会発表論文集(Web) 26th 2020年

    ISSN:2188-4420

  49. マスク言語モデルを利用したデータ拡張に基づく日本語文内ゼロ照応解析

    今野颯人, 松林優一郎, 松林優一郎, 清野舜, 清野舜, 大内啓樹, 大内啓樹, 高橋諒, 高橋諒, 乾健太郎, 乾健太郎

    言語処理学会年次大会発表論文集(Web) 26th 2020年

    ISSN:2188-4420

  50. 記述式答案自動採点のための確信度推定手法の検討

    舟山弘晃, 舟山弘晃, 佐々木翔大, 佐々木翔大, 水本智也, 水本智也, 三田雅人, 三田雅人, 鈴木潤, 鈴木潤, 松林優一郎, 松林優一郎, 乾健太郎, 乾健太郎

    言語処理学会年次大会発表論文集(Web) 26th 2020年

    ISSN:2188-4420

  51. 解説文生成研究のためのライティング技術解説付き学習者コーパス

    永田亮, 永田亮, 永田亮, 石川慎一郎, 乾健太郎, 乾健太郎

    言語処理学会年次大会発表論文集(Web) 25th 2019年

    ISSN:2188-4420

  52. 負例を厳選した対話応答選択テストセット構築の試みと分析

    佐藤志貴, 赤間怜奈, 赤間怜奈, 大内啓樹, 大内啓樹, 鈴木潤, 鈴木潤, 乾健太郎, 乾健太郎

    人工知能学会言語・音声理解と対話処理研究会資料 87th 2019年

    ISSN:0918-5682

  53. クラウドソーシングによる日本語FrameNetと自動構築した格フレームとの対応付け

    河原 大輔, 小原 京子, 関根 聡, 乾 健太郎

    言語処理学会第24回年次大会 706-709 2018年3月

  54. ニューラルヘッドライン生成における誤生成問題の改善

    清野舜, 高瀬翔, 鈴木潤, 岡崎直観, 乾健太郎, 乾健太郎, 永田昌明

    言語処理学会年次大会発表論文集(Web) 24th 2018年

    ISSN:2188-4420

  55. 複単語表現を利用した因果関係推定モデルの改善

    佐々木翔大, 高瀬翔, 高瀬翔, 井之上直也, 岡崎直観, 乾健太郎

    情報処理学会研究報告(Web) 2017 (SLP-116) Vol.2017‐SLP‐116,No.22,1‐6 (WEB ONLY) 2017年5月8日

  56. Supervised Approaches for Japanese Wikification

    Shuangshuang Zhou, Naoaki Okazaki, Koji Matsuda, Ran Tian, Kentaro Inui

    情報処理学会論文誌データベース(TOD) 10 (1) 2017年3月22日

    ISSN:1882-7799

    詳細を見る 詳細を閉じる

    Wikification is the task of connecting mentions in texts to entities in a large-scale knowledge base, Wikipedia. In this paper, we present a pipeline system for Japanese Wikification that consists of two components, namely candidate generation and candidate ranking. We investigate several techniques for each component, using a recently developed Japanese Wikification corpus. For candidate generation, we find that a name dictionary using anchor texts of Wikipedia is more effective than other methods based on similarity of surface forms. For candidate ranking, we verify that a set of features used in English Wikification is effective in Japanese Wikification as well. In addition, by using a corpus that links mentions to Japanese Wikipedia entries instead of to English Wikipedia entries, we are able to acquire rich contextual information from Japanese Wikipedia articles, which leads to improvements for Japanese mention disambiguation. We take this advantage by exploring several embedding models that encode context information of Wikipedia entities. The experimental results demonstrate that they improve candidate ranking. We also report the effect of each feature in detail. To sum, our system achieves 81.60% accuracy, significantly outperforming the previous work.------------------------------This is a preprint of an article intended for publication Journal ofInformation Processing(JIP). This preprint should not be cited. Thisarticle should be cited as: Journal of Information Processing Vol.25(2017) (online)------------------------------Wikification is the task of connecting mentions in texts to entities in a large-scale knowledge base, Wikipedia. In this paper, we present a pipeline system for Japanese Wikification that consists of two components, namely candidate generation and candidate ranking. We investigate several techniques for each component, using a recently developed Japanese Wikification corpus. For candidate generation, we find that a name dictionary using anchor texts of Wikipedia is more effective than other methods based on similarity of surface forms. For candidate ranking, we verify that a set of features used in English Wikification is effective in Japanese Wikification as well. In addition, by using a corpus that links mentions to Japanese Wikipedia entries instead of to English Wikipedia entries, we are able to acquire rich contextual information from Japanese Wikipedia articles, which leads to improvements for Japanese mention disambiguation. We take this advantage by exploring several embedding models that encode context information of Wikipedia entities. The experimental results demonstrate that they improve candidate ranking. We also report the effect of each feature in detail. To sum, our system achieves 81.60% accuracy, significantly outperforming the previous work.------------------------------This is a preprint of an article intended for publication Journal ofInformation Processing(JIP). This preprint should not be cited. Thisarticle should be cited as: Journal of Information Processing Vol.25(2017) (online)------------------------------

  57. 危機対応におけるノウハウ抽出方法の提案

    上江洲愛, 爰川知宏, 小阪尚子, 乾健太郎, 山口健史, 前田裕二

    電子情報通信学会技術研究報告 116 (488(LOIS2016 62-103)) 2017年

    ISSN:0913-5685

  58. 含意関係認識

    乾 健太郎

    人工知能学大辞典 644-645 2017年

  59. Geographical Entity Annotated Corpus of Japanese Microblogs

    Koji Matsuda, Akira Sasaki, Naoaki Okazaki, Kentaro Inui

    情報処理学会論文誌データベース(TOD) 9 (4) 2016年12月22日

    ISSN:1882-7799

    詳細を見る 詳細を閉じる

    This paper addresses the issues in the task of annotating geographical entities on microblogs and reports the preliminary results of our efforts to annotate Japanese microblog texts. Unlike prior work, we aim at annotating not only geographical location entities but also facility entities, such as stations, restaurants and schools. We discuss (i) how to build a gazetteer of geographical entities with a sufficiently broad coverage, (ii) what types ambiguities that need to be considered, (iii) why the annotator tends to disagree, and (iv) what technical problems should be addressed to automate the task of annotating the geographical entities. All the annotation data and the annotation guidelines are publicly available for research purposes from our web site.------------------------------This is a preprint of an article intended for publication Journal ofInformation Processing(JIP). This preprint should not be cited. Thisarticle should be cited as: Journal of Information Processing Vol.25(2017) (online)------------------------------This paper addresses the issues in the task of annotating geographical entities on microblogs and reports the preliminary results of our efforts to annotate Japanese microblog texts. Unlike prior work, we aim at annotating not only geographical location entities but also facility entities, such as stations, restaurants and schools. We discuss (i) how to build a gazetteer of geographical entities with a sufficiently broad coverage, (ii) what types ambiguities that need to be considered, (iii) why the annotator tends to disagree, and (iv) what technical problems should be addressed to automate the task of annotating the geographical entities. All the annotation data and the annotation guidelines are publicly available for research purposes from our web site.------------------------------This is a preprint of an article intended for publication Journal ofInformation Processing(JIP). This preprint should not be cited. Thisarticle should be cited as: Journal of Information Processing Vol.25(2017) (online)------------------------------

  60. 災害情報システムでのタスク進捗管理における標準処理手順(SOP)策定の有効性評価

    小阪尚子, 小山晃, 爰川知宏, 前田裕二, 河本尋子, 鈴木進吾, 山口健史, 乾健太郎

    地域安全学会梗概集(CD-ROM) (39) 2016年

  61. 交通オントロジーと説明生成に基づく交通危険予測

    高橋 諒, 井之上 直也, 栗谷 康隆, 山本 風人, 乾 健太郎

    JSAI大会論文集 2016 (0) 4I41-4I41 2016年

    出版者・発行元:一般社団法人 人工知能学会

  62. 自然言語処理技術の概要

    乾健太郎, 関根聡

    情報処理 57 (1) 6-9 2016年1月

  63. Project Next NLP 〜エラー分析を通じた自然言語処理技術の推進〜

    関根聡, 乾健太郎

    情報処理 57 (1) 3-5 2016年1月

  64. 自然言語処理技術の現状と展望 -エラー分析プロジェクトを通して-:0.編集にあたって

    関根 聡, 乾 健太郎

    情報処理 57 (1) 2-2 2015年12月15日

  65. 自然言語処理

    乾健太郎

    この1冊でまるごとわかる人工知能ビジネス 2015年8月

    出版者・発行元:日経BP社

  66. 対数的共起ベクトルの加法構成性

    田 然, 岡崎 直観, 乾 健太郎

    情報処理学会研究報告. SLP, 音声言語情報処理 2015 (14) 1-12 2015年5月18日

    出版者・発行元:一般社団法人情報処理学会

    詳細を見る 詳細を閉じる

    この論文では,単語ベクトルの算術平均によって短いフレーズの意味を近似できる理由について初めての数学的解明を行う.具体的には,その近似による 「誤差」 に対する上界が理論的に与えられ,実験的に検証された.このような加法構成性が成り立つ必要条件として,対数関数と文脈のオーバーラップが重要であることや,低い共起頻度を Zipf 則に従って補完するのが有効であることなど,理論上予測される幾つかの性質も実験によって確かめられた.更に,加法構成性を考える上では,特異値分解による単語埋め込みは,最先端な埋め込み手法に匹敵する性能を達成できることを示す.

  67. 単語分散表現のshift-reduce型構文解析への利用

    小松 広弥, 田 然, 岡崎 直観, 乾 健太郎

    情報処理学会研究報告. SLP, 音声言語情報処理 2015 (3) 1-8 2015年5月18日

    出版者・発行元:一般社団法人情報処理学会

    詳細を見る 詳細を閉じる

    本研究では,英語の shift-reduce 型依存構文解析器において,単語の分散表現を素性に利用することで,単語の意味,構文構造的なクラスを捉え,解析の精度が向上することを示す.これは,類似する単語は,その単語に関する依存構文が類似しているという考えに基づく.単語分散表現の構築については,大量の言語データから分布意味論仮説に基づき,周辺単語の統計情報を利用するような一般的な構築手法に加え,解析器の内部状態を利用し,解析器の内部動作に着目した単語分散表現の構築手法を提案する.

  68. 物理モデルと論理推論の統合による運転シーンの潜在的危険の予測

    小林颯介, 井之上直也, 栗谷康隆, 乾健太郎, 近藤敏之, 安部克則, 奥野英一

    自動車技術会大会学術講演会講演予稿集(CD-ROM) 2015 ROMBUNNO.S205 2015年5月18日

    ISSN:2189-4558

  69. 場所参照表現タグ付きコーパスの構築と評価

    松田 耕史, 佐々木 彬, 岡崎 直観, 乾 健太郎

    研究報告自然言語処理(NL) 2015 (12) 1-10 2015年1月12日

    出版者・発行元:一般社団法人情報処理学会

    詳細を見る 詳細を閉じる

    ソーシャルメディア上からサンプリングした日本語テキストに中に現れる場所参照表現に対して座標情報を含む具体的なエンティティ情報を付与したコーパスを試作した.その際,エンティティ辞書をどのように構築するか,アノテーション付与の対象をどの範囲にするかなど,様々な設計上の選択を行うことが必要であったため,この過程で行われた議論について述べる.また,ガイドラインの妥当性を検証するために複数人でアノテーションを行い,その結果浮かび上がった課題を整理し,実際に場所参照表現のグラウンディングを行うためにはどのような技術的課題を解決しなければならないかを考察する.

  70. 関係知識獲得のための意味表現の学習

    高瀬 翔, 岡崎 直観, 乾 健太郎

    人工知能学会全国大会論文集 29 1-4 2015年

    出版者・発行元:人工知能学会

    ISSN:1347-9881

  71. 言語処理のための仮説推論エンジンPhillip

    山本風人, 井之上直也, 乾健太郎

    言語処理学会年次大会発表論文集(Web) 21st C2-3 (WEB ONLY) 2015年

  72. 《特別鼎談》編集長3人は語り尽くせなかった

    平田圭二, 塚本昌彦, 乾健太郎

    情報処理 56 (1) 86-92 2014年12月15日

  73. 言葉が分かるコンピューターを作る

    乾健太郎

    「科学の泉」 2014年12月

    出版者・発行元:河北新報

  74. 共参照解析のための事象間関係知識の文脈化

    井之上直也, 杉浦純, 乾健太郎

    計測自動制御学会システム・情報部門学術講演会講演論文集(CD-ROM) 2014 ROMBUNNO.SS2-8 2014年11月21日

  75. 近似頻度計測手法を用いた大規模データからの関係獲得

    高瀬翔, 岡崎直観, 乾健太郎

    研究報告自然言語処理(NL) 2014 (9) 1-9 2014年6月26日

    出版者・発行元:一般社団法人情報処理学会

    詳細を見る 詳細を閉じる

    大規模な Web データからエンティティペアやパタンのクラスタリングによって意味関係を抽出する教師なし関係抽出タスクにおいて,パタンやエンティティペアの間の類似度計算をスケーラブルに行う事は非常に重要な問題である.本論文では,近似頻度計算による省メモリな素性作成手法と次元圧縮による類似度計算時間の高速化を実現する手法を提案する.実験において,近似計算でも精度が落ちない事,また,計算時間の高速化の効果を示す.

  76. Word2vecの並列実行時の学習速度の改善

    岡崎直観, 乾健太郎

    研究報告自然言語処理(NL) 2014 (8) 1-5 2014年6月26日

    出版者・発行元:一般社団法人情報処理学会

    詳細を見る 詳細を閉じる

    単語の意味ベクトルを大規模コーパスから学習するためのツールとして,Mikolov らの手法 [14] を実装した word2vec が注目を浴びている.本論文は,word2vec を複数のプロセッサで並列で動作させた時に学習速度が低下する原因を説明し,これを改善するアルゴリズムを提案する.提案手法は学習で得られる単語ベクトルの質を落とすこと無く,複数のプロセッサを効率よく利用できることを実験的に示す.

  77. A<sup>*</sup>探索に基づく仮説推論の効率化

    山本 風人, 井之上 直也, 乾 健太郎, 荒瀬 由紀, 辻井 潤一

    研究報告自然言語処理(NL) 2014 (10) 1-9 2014年6月26日

    出版者・発行元:一般社団法人情報処理学会

    詳細を見る 詳細を閉じる

    仮説推論は、観測に対する最良の説明を求める推論である。仮説推論は、文章に明示されていない情報の顕在化を行うための有望な枠組みとして考えられてきた一方で、最良の説明を求めるための計算時間が入力の規模に対して指数関数的に増加するため、実問題への適用事例は未だ限定的であるのが現状である。そこで我々は、仮説推論の探索空間を効率良く削減するために、A* 探索に基づく説明候補生成アルゴリズムを提案する。より具体的には、複数の証拠により支持されない仮説リテラルを、効率良く探索空間から排除する手法を提案する。また、評価実験を通して、最新の仮説推論アルゴリズムの計算効率が提案手法により大幅に改善されることを報告する。Abduction is inference to the best explanation. While abduction has long been considered as a promising framework for natural language processing (NLP), its computational complexity hinders applying abduction to practical NLP problems. In this paper, we adapt an A* search-based technique for abduction in order to prune its search space efficiently. More specifically, we present an algorithm to effectively eliminate literals which do not introduce any unification from the search space. Our experiment reveals that our search-based technique considerably reduces the size of the search space and improves the efficiency of the state-of-the-art inference algorithm for abduction on first-order logic.

  78. 述部機能表現に対する意味ラベル付与

    上岡 裕大, 成田 和弥, 水野 淳太, 乾 健太郎

    研究報告自然言語処理(NL) 2014 (9) 1-9 2014年5月15日

    出版者・発行元:一般社団法人情報処理学会

    詳細を見る 詳細を閉じる

    機能表現を正しく認識することは,拡張モダリティ解析や事実性解析をはじめとする応用タスクに有用である.しかし,解析ツールを開発するための基礎となるコーパスが利用可能でないことや,機能表現解析に関わる問題が明らかでないなどの課題がある.そこで,本研究では,応用タスクとして日本語事実性解析を例に挙げ,その精度向上を目的として既存の機能表現辞書の再構築,機能表現意味ラベル付与コーパスの作成および,そのコーパスを用いた機能表現解析を行った.Semantic tagging of functional expressions is useful for a broad range of NLP applications such as extended modality analysis and factuality analysis. However, automatic semantic tagging of functional expressions has some technical challenges. For example, the corpus of functional expressions is unavailable. In this paper, for improvement of factuality analysis, we present three contributions; (i) a functional expression dictionary based on the existing dictionary, (ii) an available corpus of functional expressions and (iii) an automatic semantic tagger for functional expressions.

  79. 述部機能表現に対する意味ラベル付与

    上岡 裕大, 成田 和弥, 水野 淳太, 乾 健太郎

    研究報告音声言語情報処理(SLP) 2014 (9) 1-9 2014年5月15日

    出版者・発行元:一般社団法人情報処理学会

    詳細を見る 詳細を閉じる

    機能表現を正しく認識することは,拡張モダリティ解析や事実性解析をはじめとする応用タスクに有用である.しかし,解析ツールを開発するための基礎となるコーパスが利用可能でないことや,機能表現解析に関わる問題が明らかでないなどの課題がある.そこで,本研究では,応用タスクとして日本語事実性解析を例に挙げ,その精度向上を目的として既存の機能表現辞書の再構築,機能表現意味ラベル付与コーパスの作成および,そのコーパスを用いた機能表現解析を行った.Semantic tagging of functional expressions is useful for a broad range of NLP applications such as extended modality analysis and factuality analysis. However, automatic semantic tagging of functional expressions has some technical challenges. For example, the corpus of functional expressions is unavailable. In this paper, for improvement of factuality analysis, we present three contributions; (i) a functional expression dictionary based on the existing dictionary, (ii) an available corpus of functional expressions and (iii) an automatic semantic tagger for functional expressions.

  80. ネットにあふれる健康・医療情報の信頼性をどう担保するか? 〜自然言語処理からの情報科学的アプローチ〜

    乾健太郎

    仙台医師会報 2014年4月

  81. ビッグデータから知をつむぐ自然言語処理

    乾健太郎

    翠巒 2014年3月

  82. 同一文内の表現対を対象とした日本語における時間関係認識

    稲田和明, 松林優一郎, 乾健太郎

    言語処理学会年次大会発表論文集(Web) 20th 2014年

    ISSN:2188-4420

  83. 共参照解析のための事象間関係知識の文脈化

    井之上直也, 杉浦純, 乾健太郎

    言語処理学会年次大会発表論文集(Web) 20th A5-3 (WEB ONLY) 2014年

  84. 共参照解析における事象間関係知識の適用

    杉浦純, 井之上直也, 乾健太郎

    言語処理学会年次大会発表論文集(Web) 20th A5-2 (WEB ONLY) 2014年

  85. 情報信頼性検証への自然言語処理的アプローチ (一般セッションおよび「コミュニケーションにおける安心・安全・信頼」に関する特別セッション)

    乾 健太郎, 岡崎 直観

    言語・音声理解と対話処理研究会 69 23-27 2013年12月22日

    出版者・発行元:人工知能学会

    ISSN:0918-5682

  86. 訂正パターンに基づく誤情報の抽出と集約

    鍋島啓太, 水野淳太, 岡崎直観, 乾健太郎

    第75回全国大会講演論文集 2013 (1) 179-180 2013年3月6日

    詳細を見る 詳細を閉じる

    東日本大震災では、誤情報の拡散が問題となった。その中には「イソジンが被曝に効果がある」のように人体に悪影響を及ぼすものも少なくない。本研究では、誤情報を訂正する表現(以下、訂正パターン)に着目し、東日本大震災後1週間の全ツイートから誤情報を網羅的に抽出することを目指す。具体的には、訂正パターンを人手で整備し、訂正パターンにマッチするツイートを抽出する。次に、収集したツイートを内容の類似性に基づいてクラスタリングし、最後に、その中から誤情報を過不足なく説明する1文を選択する。実験では、誤情報を人手でまとめたウェブサイトを正解データとして、評価を行った。

  87. 返信・非公式リツイートで表明される論述関係の認識

    大和田裕亮, 岡崎直観, 乾健太郎, 石塚満

    第75回全国大会講演論文集 2013 (1) 171-172 2013年3月6日

    詳細を見る 詳細を閉じる

    東日本大震災では安否確認や被災者支援のためにTwitterが活躍したが,一方で多種多様な情報が流通し,混乱を招いた.我々は,情報の信憑性や重要性を評価するには,ツイート空間の論述的な構造を解析・可視化し,情報の「裏」を取ることが大切だと考えている.本稿では,ツイートの返信および非公式リツイートによって投稿者の「同意」「反論」「疑問」などの態度が表明されると考え,これらを推定する分類器を教師有り学習で構築する.評価実験では,東日本大震災時のツイート群を対象とし,返信・非公式リツイートで表明される態度を推定した時の性能を報告する.

  88. Twitterにおける誤情報の拡散収束過程の可視化

    渡邉研斗, 鍋島啓太, 水野淳太, 岡崎直観, 乾健太郎

    全国大会講演論文集 2013 (1) 657-659 2013年3月6日

    出版者・発行元:一般社団法人情報処理学会

    詳細を見る 詳細を閉じる

    東日本大震災時では、電話やメールよりも大量の情報を即座に得ることができるTwitterが情報提供・収集に大きく貢献した。一方で、「コスモ石油の爆発で有害な雨が降る」などの誤情報が拡散し、人々の混乱の要因となった。本研究では、東日本大震災後1週間の全ツイートを用い、誤情報の発生・拡散・収束までの過程を分析・可視化する。この分析を正確に行うため、誤情報を拡散するツイートと訂正するツイートを手作業で分類した。さらに、この作業を自然言語処理技術で自動的に行う手法を提案し、誤情報に対してその訂正情報を自動的に提示するシステムの性能や今後の課題を報告する。

  89. マイクロブログユーザからの現地被災者抽出の技術的支援

    水野淳太, 岡崎直観, 乾健太郎

    全国大会講演論文集 2013 (1) 523-525 2013年3月6日

    出版者・発行元:一般社団法人情報処理学会

    詳細を見る 詳細を閉じる

    災害時にマイクロブログに投稿された情報は,来たるべき災害に備えるために有益であると考えられている.中でも,被災者が発信した情報は,今災害時の問題点を明らかにし,今後の対策を講じるために重要な役割を果たす.そのため,多くの情報発信者の中から被災者を同定する技術は重要である.本研究では,投稿情報中の地名や画像データに着目して,実際に災害を経験した被災者を抽出する手法を提案する.地名は,情報の発信場所や被害のあった場所の情報などを示している可能性が高い.画像は,それが被災地で撮影された写真ならば,被災者の撮影したものと考えられる.発表では,それらの情報から被災者をどの程度同定できたかについて報告する.

  90. マイクロブログにおける同意・反論関係を用いた情報信頼度推定

    佐藤 雅宏, 岡崎 直観, 乾 健太郎

    人工知能学会全国大会論文集 27 1-4 2013年

    出版者・発行元:人工知能学会

    ISSN:1347-9881

  91. ウェブ文書の構造を利用した場所名・住所ペアの獲得

    佐藤 貴大, 岡崎 直観, 乾 健太郎

    人工知能学会全国大会論文集 27 1-4 2013年

    出版者・発行元:人工知能学会

    ISSN:1347-9881

  92. Discriminative learning of first-order weighted abduction from partial discourse explanations

    Kazeto Yamamoto, Naoya Inoue, Yotaro Watanabe, Naoaki Okazaki, Kentaro Inui

    Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) 7816 (1) 545-558 2013年

    DOI: 10.1007/978-3-642-37247-6_44  

    ISSN:0302-9743 1611-3349

    詳細を見る 詳細を閉じる

    Abduction is inference to the best explanation. Abduction has long been studied in a wide range of contexts and is widely used for modeling artificial intelligence systems, such as diagnostic systems and plan recognition systems. Recent advances in the techniques of automatic world knowledge acquisition and inference technique warrant applying abduction with large knowledge bases to real-life problems. However, less attention has been paid to how to automatically learn score functions, which rank candidate explanations in order of their plausibility. In this paper, we propose a novel approach for learning the score function of first-order logic-based weighted abduction [1] in a supervised manner. Because the manual annotation of abductive explanations (i.e. a set of literals that explains observations) is a time-consuming task in many cases, we propose a framework to learn the score function from partially annotated abductive explanations (i.e. a subset of those literals). More specifically, we assume that we apply abduction to a specific task, where a subset of the best explanation is associated with output labels, and the rest are regarded as hidden variables. We then formulate the learning problem as a task of discriminative structured learning with hidden variables. Our experiments show that our framework successfully reduces the loss in each iteration on a plan recognition dataset. © 2013 Springer-Verlag.

  93. Evidence in automatic error correction improves learners' English skill

    Jiro Umezawa, Junta Mizuno, Naoaki Okazaki, Kentaro Inui

    Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) 7817 (2) 559-571 2013年

    DOI: 10.1007/978-3-642-37256-8_46  

    ISSN:0302-9743 1611-3349

    詳細を見る 詳細を閉じる

    Mastering proper article usage, especially in the English language, has been known to pose an extreme challenge to non-native speakers whose L1 languages have no concept of articles. Although the development of correction methods for article usage has posed a challenge for researchers, current methods do not perfectly correct the articles. In addition, proper article usage is not taught by these methods. Therefore, they are not useful for those wishing to learn a language with article usage. In this paper, we discuss the necessity of presenting evidence for corrections of English article usage. We demonstrate the effectiveness of this approach to improve the writing skills of English learners. © 2013 Springer-Verlag.

  94. 言語処理による分析-日本栄養士会活動報告の分析

    岡崎直観, 鍋島啓太, 乾健太郎

    日本栄養士会雑誌 55 (12) 6-8 2012年12月

  95. Online Large-margin Weight Learning for First-order Logic-based Abduction (情報論的学習理論と機械学習・第15回情報論的学習理論ワークショップ)

    INOUE Naoya, YAMAMOTO Kazeto, WATANABE Yotaro, OKAZAKI Naoaki, INUI Kentaro

    電子情報通信学会技術研究報告. IBISML, 情報論的学習理論と機械学習 = IEICE technical report. IBISML, Information-based induction sciences and machine learning 112 (279) 143-150 2012年10月31日

    出版者・発行元:一般社団法人電子情報通信学会

    ISSN:0913-5685

    詳細を見る 詳細を閉じる

    Abduction is inference to the best explanation. Abduction has long been studied in a wide range of contexts and is used for modeling artificial intelligence systems, such as diagnostic systems and plan recognition systems. However, less attention has been paid to how to automatically learn score functions, which rank explanations in the order of their plausibility. In this paper, we propose a supervised learning approach for first-order logic-based abduction. The contribution of this paper is the following: (i) we show how to formulate the machine learning problem of abduction with the framework of online large-margin training, which has been shown to have both predictive performance and scalability to larger problems; (ii) we extend the state-of-the-art abductive reasoning system [15] to model the score function with a weighted linear model, which is the groundwork for the online large-margin training; (iii) we support partially-specified gold-standard explanations as training examples, where the weights are learned to rank any explanation that includes the gold-standard explanation as the best explanation; (iv) the all-in-one software package for inference and learning is made publicly available.

  96. Extending ILP-based Abductive Reasoning with Cutting Plane Inference

    INOUE Naoya, INUI Kentaro

    情報処理学会研究報告(CD-ROM) 2012 (3) ROMBUNNO.NL-208,NO.5-121 2012年10月15日

    DOI: 10.3233/WIA-2011-0209  

    ISSN:2186-2583

  97. 誤差逆伝播を利用した重み付き仮説推論の教師あり学習

    山本 風人, 井之上 直也, 渡邊 陽太郎, 岡崎 直観, 乾 健太郎

    研究報告自然言語処理(NL) 2012 (9) 1-8 2012年5月3日

    出版者・発行元:一般社団法人情報処理学会

    詳細を見る 詳細を閉じる

    本稿では、重み付き仮説推論のパラメタを教師あり学習によって自動調整する手法を提案する。仮説推論は、与えられた観測に対して評価関数を最大化する最良の説明を求める推論であり、自然言語処理において文章に明示されていない情報の顕在化を行うことに対して有用な枠組みとして注目を浴びている。しかしその一方で、仮説推論の評価関数の学習手法が未だ提案されておらず、評価関数のパラメタの調整は、人手による調整やヒューリスティックな手法に頼らざるを得ないという問題があった。そこで我々は、仮説推論の拡張のひとつである重み付き仮説推論を対象として、仮説の証明木におけるリテラル間のコストの関係をフィードフォワードニューラルネットワークの形で表現することで、誤差に対する各パラメタの勾配を求め、評価関数のパラメタの識別学習を実現する。また、提案手法によって評価関数を学習できていることを確かめるために、既存のデータセットを用いて実験した結果についても報告する。We explore a discourse processing framework for discovering implicit information in texts, based on Hobbs et al.&#039;s weighted abduction [5]. Abduction is inference to the best explanation. In weighted abduction, the best explanation is defined as the explanation that minimizes a parametrized cost function. However, less attention has been paid to how to tune the parameters of the cost function automatically. In this paper, we propose a discriminative approach to learning parameters in weighted abduction. We represent the transition of costs in a proof tree as feed-forward neural networks, and calculate the gradients of parameters in a background knowledge base. Our experiments show that our method correctly learns parameters on the existing dataset of plan recognition.

  98. Natural Logicと条件付確率場の融合による構成的文間関係認識

    渡邉陽太郎, 水野淳太, 岡崎直観, 乾健太郎

    情報処理学会自然言語処理研究会予稿集, NL-208-6 2012年

  99. Compositional Semantic Relation Recognition between Sentences by Combining Conditional Random Fields and Natural Logic

    渡邉陽太郎, 水野淳太, 岡崎直観, 乾健太郎

    情報処理学会自然言語処理研究会予稿集, NL-208-6 2012年

  100. Organizing information on the web through agreement-conflict relation classification

    Junta Mizuno, Eric Nichols, Yotaro Watanabe, Kentaro Inui

    Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) 7675 126-137 2012年

    DOI: 10.1007/978-3-642-35341-3_11  

    ISSN:0302-9743 1611-3349

    詳細を見る 詳細を閉じる

    The vast amount of information on the Web makes it difficult for users to comprehensively survey the various viewpoints on topics of interest. To help users cope with this information overload, we have developed an Information Organization System that applies state-of-theart technology from Recognizing Textual Entailment to automatically detect Web texts that are relevant to natural language queries and organize them into agreeing and conflicting groups. Users are presented with a bird's-eye-view visualization of the viewpoints on their queries that makes it easier to gain a deeper understanding of an issue. In this paper, we describe the implementation of our Information Organization System and evaluate our system through empirical analysis of the semantic relation recognition system that classifies texts and through a large-scale usability study. The empirical evaluation and usability study both demonstrate the usefulness of our system. User feedback further shows that by exposing our users to differing viewpoints promotes objective thinking and helps to reduce confirmation bias. © Springer-Verlag 2012.

  101. Large-scale cost-based abduction in full-fledged first-order predicate logic with cutting plane inference

    Naoya Inoue, Kentaro Inui

    Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) 7519 281-293 2012年

    DOI: 10.1007/978-3-642-33353-8_22  

    ISSN:0302-9743 1611-3349

    詳細を見る 詳細を閉じる

    Abduction is inference to the best explanation. Abduction has long been studied intensively in a wide range of contexts, from artificial intelligence research to cognitive science. While recent advances in large-scale knowledge acquisition warrant applying abduction with large knowledge bases to real-life problems, as of yet no existing approach to abduction has achieved both the efficiency and formal expressiveness necessary to be a practical solution for large-scale reasoning on real-life problems. The contributions of our work are the following: (i) we reformulate abduction as an Integer Linear Programming (ILP) optimization problem, providing full support for first-order predicate logic (FOPL) (ii) we employ Cutting Plane Inference, which is an iterative optimization strategy developed in Operations Research for making abductive reasoning in full-fledged FOPL tractable, showing its efficiency on a real-life dataset (iii) the abductive inference engine presented in this paper is made publicly available. © 2012 Springer-Verlag.

  102. 日本語事実性解析課題の経験的分析

    成田 和弥, 水野 淳太, 乾 健太郎

    研究報告自然言語処理(NL) 2011 (17) 1-8 2011年11月14日

    詳細を見る 詳細を閉じる

    事実性とは,文中のある事象が実際に起こったことなのか,あるいは起こる可能性を述べただけなのか,に関する情報である.このような事実性の解析は,情報抽出や含意関係認識などに有用であるが,内在する問題の分析・整理が不十分であり,いまだ十分な解析性能は実現されていない.本稿では,高度な日本語事実性解析器の実現に向けて,事実性に影響を与える手がかりとなる表現に着目し,それら語彙知識の組み合わせによって,どの程度の事実性解析課題を解決できるのかを分析した.具体的には,事象の成立に影響を与える表現を手がかりとして,事実性の解析を行うSauri (2008) の解析モデルをもとに,日本語事実性解析器を構築した.そして,構築した解析器をウェブ上の数千文に対して適用し,誤り分析を行った.Event factuality is information about whether events mentioned in natural language correspond to actual events that have occurred in the real world or to events that are of uncertain interpretation. Factuality analysis is useful for information extraction and recognizing textual entailment, among others, but the problems inherent to factuality analysis are not yet fully-understood, and sufficient performance has not yet been realized. In this paper, we make efforts toward a high-performance Japanese factuality analyzer by focusing on "factuality markers" (Sauri 2008), expressions that can influence the factuality of events, and determine the level of performance that can be obtained through a compositional approach using these markers. We construct a Japanese factuality analyzer following the approach of Sauri (2008) and conduct error analysis on several thousand sentences from the Web.

  103. An ILP Formulation of Abductive Inference for Discourse Interpretation

    INOUE Naoya, INUI Kentaro

    情報処理学会研究報告(CD-ROM) 2011 (3) ROMBUNNO.NL-203,NO.3 2011年10月15日

    ISSN:2186-2583

  104. 情報爆発時代の Web のサーチ・分析技術

    黒橋 禎夫, 乾 健太郎

    電子情報通信学会誌 = The journal of the Institute of Electronics, Information and Communication Engineers 94 (8) 673-678 2011年8月1日

    出版者・発行元:一般社団法人電子情報通信学会

    ISSN:0913-5693

    詳細を見る 詳細を閉じる

    近年のWebの爆発的な広がりは,「情報爆発」と呼ぶにふさわしいものであり,膨大なWeb空間の中にいわば森羅万象の情報が散在している.膨大な量のWebページから求めるページを選択するためのサーチ技術,また,膨大な情報を分析し有用な情報を抽出するための分析技術が極めて重要となっている.本稿では,これらのトピックについて,深い言語処理に基づく大規模知識空間のサーチ・分析技術に焦点を当てて,背景と現状の技術,及び今後の展望について述べる.

  105. 情報分析システム WISDOM のユーザ評価とその分析

    川田 拓也, 赤峯 享, 河原 大輔, 加藤 義清, 乾 健太郎, 黒橋 禎夫, 木俵 豊

    言語処理学会 第17回年次大会 45-48 2011年3月

  106. 大規模Web情報分析のための分析対象ページの段階的選択

    赤峯 享, 加藤 義清, 川田 拓也, レオン 末松豊インティ, 河原 大輔, 乾 健太郎, 黒橋 禎夫

    言語処理学会 第17回年次大会 41-44 2011年3月

  107. 競合事物間における比較関係認識

    山崎 義隆, 乾 健太郎, 松本 裕治

    情報処理学会研究報告. 自然言語処理研究会報告 2011 (5) 1-7 2011年1月21日

    出版者・発行元:情報処理学会

    ISSN:1884-0930

    詳細を見る 詳細を閉じる

    人々が,ある商品やサービスについてウェブ上で評判情報を探す場合,そのものに関する情報を収集するだけではなく,たいてい,競合する商品やサービスについての評判情報も収集する.この時,当初の目的の事物とそれと競合する事物を比較した意見や感想は,非常に有益である.なぜなら,このような競合事物の比較文から,目的の事物の優位点や欠点,相違点等を明示的に知ることができるからである.本稿では,ユーザーが入力した競合する 2 つの事物に対して,それらを明示的に比較した比較文をテキスト集合から抽出し,その文において競合事物対がどのような比較関係 (有差,同等,最上級,特徴) にあるか判定する手法を提案する.提案手法では,日本語の比較表現を観察することにより人手で構築した規則集合と,教師あり機械学習手法を併用する.9 組の競合事物対を用いて提案手法の評価実験を行い,その有効性を確かめた.

  108. Organizing Agreeing and Con icting Opinions based on Semantic Relation Recognition

    Junta Mizuno, Yotaro Watanabe, Eric Nichols, Koji Murakami, Kentaro Inui, Yuji Matsumoto

    情報処理学会論文誌 52 (12) 2011年

  109. Web時代の音声・言語技術

    中村哲, 磯谷亮輔, 乾健太郎, 柏岡秀紀, 河井恒, 河原達也, 木俵豊, 黒橋禎夫, 隅田英一郎, 関根聡, 鳥澤健太郎, 堀智織, 松田繁樹

    電子情報通信学会誌 94 (6) 2011年

  110. 文間の弱い対立関係の認識

    大木 環美, 村上 浩司, 松吉 俊, 水野 淳太, 乾 健太郎, 松本 裕治

    情報処理学会研究報告. 自然言語処理研究会報告 2010 (7) 1-9 2010年11月11日

    出版者・発行元:情報処理学会

    ISSN:1884-0930

    詳細を見る 詳細を閉じる

    意見マイニング等の意見を集約するようなタスクでは,賛成意見や反対意見を分類することが行われている.実際には,これらの意見の中には部分的に賛成意見のものや部分的に反対意見のものが存在しており,完全な賛成意見や完全な反対意見とは異なる観点からの有用な情報を述べている.これらを認識するために有用な含意関係認識 (RTE) のタスクでは,文対は含意・矛盾・不明の 3 種類の関係に分類される.しかし,現状ではまったくの賛成意見や反対意見と区別されず同等として扱われてしまうため,ユーザが知見としてそれらの情報を得ることが難しい.本稿では,このような有用な情報を認識するための意味的関係を弱対立の関係と呼ぶ。この弱対立関係を新たに定義し,与えられた 2 文間の関係が弱対立関係であるかを認識するための弱対立認識の手がかかりを作成する.さらにその手がかりを用いたルールベースのシステムを実装し,弱対立認識の評価と現状の問題点を示す.

  111. 大規模言語/世界知識に基づく次世代 Web 情報分析

    乾 健太郎

    電子情報通信学会技術研究報告. COMP, コンピュテーション 110 (232) 17-17 2010年10月8日

    出版者・発行元:一般社団法人電子情報通信学会

  112. 帰属文書数に基づくWebページ情報発信者の専門性分析

    加藤 義清, 乾 健太郎, 黒橋 禎夫

    研究報告情報基礎とアクセス技術(IFAT) 2010 (3) 1-6 2010年7月28日

    詳細を見る 詳細を閉じる

    本研究では Web ページ情報発信者の任意のトピックにおける専門性を分析する方法として,検索エンジンのヒット数に基づき発信者の専門性スコアを計算する手法と,発信者に帰属する Web ページ数に基づき発信者の専門性スコアを計算する手法を提案する.1 億 2 千万件の日本語 Web ページを検索対象とする収集検索基盤を用いた評価実験を実施した結果,帰属文書数に基づく手法が精度や処理時間について優れていることが分かった.In this study, we propose two methods that analyze the expertise of information senders of Web pages: 1) a method that computes expertise score based on hit count from a search engine (hit count method), and 2) a method that computes expertise score based on the number of documents that are attributed to an information sender (attribution count method). We evalated both methods using a crawl and search infrastrucure which indexes 120 million Japanese Web pages. The results show that the attribution count method outperforms hit count method in terms of precision and processing time.

  113. テキスト情報分析のための判断情報アノテーション(自然言語処理,意味解析,<特集>情報爆発論文)

    松吉 俊, 江口 萌, 佐尾 ちとせ, 村上 浩司, 乾 健太郎, 松本 裕治

    電子情報通信学会論文誌. D, 情報・システム 93 (6) 705-713 2010年6月1日

    出版者・発行元:一般社団法人電子情報通信学会

    ISSN:1880-4535

    詳細を見る 詳細を閉じる

    ブログに代表される個人型情報発信メディアの爆発的な普及に伴い,膨大なテキスト情報がWeb上に加速度的に蓄積されつつある.これらの情報を目的に合わせてうまく整理し,そこから有益なデータを得るためには,記述される個々の事象に対して,その述語と項構造(だれが何をするのか)を認識するだけではなく,書き手が表明している態度や真偽判断,価値判断などの情報も解析し,総合的に分析を行う必要がある.我々は,書き手の態度や真偽判断などに関する情報を事象のモダリティ情報と呼び,文章に存在する各事象に対して,このモダリティ情報を解析する技術の開発に取り組んでいる.本論文では,事象のモダリティ情報を表す七つ組のタグの体系を提案し,現在構築しているコーパスについて報告する.また,提案するタグ体系に基づく解析システムの実現可能性とその性能に関して議論する.コーパスの対象は約4万の事象であり,タグ付与の作業者間一致度はkappa統計量で0.71という高い値であった.我々が試作した解析システムは,真偽判断タグに関して正解率0.95,F値平均0.68という,自動タグ付与機構の実現可能性を示唆する比較的高い値を示した.

  114. 文間関係認識のための局所構造アライメント

    水野 淳太, 後藤 隼人, 渡邉 陽太郎, 村上 浩司, 乾 健太郎, 松本 裕治

    情報処理学会研究報告. SLP, 音声言語情報処理 81 (16) P1-P8 2010年5月27日

    出版者・発行元:情報処理学会

    ISSN:1884-0930

    詳細を見る 詳細を閉じる

    一対の文が与えられたときに,その間にある意味的な関係を認識することは文間関係認識と呼ばれ,情報アクセス技術にとって重要である.近年では,1) 基本的な解析,2) 文間で対応する単語間に対応付け (アライメント),3) 関係分類という流れで構成される手法が広く研究されている.しかしながら,単語レベルでは対応がとれても,単語間の意味的なつながりのレベルでは対応させてはいけない場合がある.そこで,一方の文における 2 単語間の意味的な関係が,他方の文において単語アライメントされる2単語間でも成り立っているかどうかを判別する問題を考え,これを局所構造アライメントと呼び,単語アライメントと関係分類の間の処理として明示的に行うことを提案する.本稿では,局所構造アライメントの定義,およびその実装・性能評価を行った.また,文間関係認識に本手法を適用することで,文間関係認識に大きく貢献することを示す.The task of recognizing semantic relations between a pair of sentences is called cross-sentential relation recognition, and it is important for information access technology. In recent years cross-sentential relation recognition methods based on word level alignment have been widely researched and generally consist of the following steps: 1) basic processing, 2) word alignment, 3) relation classification. However, even if it is possible to align sentences at the word level, there are cases where sentences should not be aligned because their meanings are not represented in the word level alignments. Therefore, we propose adding an explicit step of processing in between word alignment and relation classification called "local structural alignment" that determines if a semantic relation between two words in one sentence is also present between the words aligned to them in the other sentence. In this paper we define local structural alignment, present an implementation, and evaluate its performance. We also show the proposed alignment method makes a large contribution to cross-sentential relation recognition.

  115. 大規模言語/世界知識に基づく Web 情報編集の高度化

    乾 健太郎

    電子情報通信学会技術研究報告. TL, 思考と言語 110 (63) 19-19 2010年5月21日

    出版者・発行元:一般社団法人電子情報通信学会

    詳細を見る 詳細を閉じる

    人間に近い高度な言語情報処理能力を工学的に実現するには,文法や辞書などの言語知識の他に,大量の世界知識を計算機に与え,言語の表層に明示される意味からその裏にある暗示的な意味を推論する頑健な機構が必要である.例えば,近年研究者の注目を集めている含意関係認識や対話・物語理解のためのプラン・意図推定などは,高度な言語処理に必要と考えられる推論の典型である.プラン推定に代表される深い言語理解の研究は1980年代に遡るが,当時の研究はそれに必要な大量の言語知識や常識的世界知識を用意する方法論を持たなかったために,実用規模には発展しなかった.これに対し,大規模テキストからの多様な知識獲得が実用性を帯びてきた現在,そこから得られる大規模な言語/世界知識を利用し,一段深い言語意味処理の実現を目指す素地が整いつつある.この講演では,講演者が現在取り組んでいる,(a)プログから意見・経験情報を抽出し,様々な意味的索引を付けてデータベース化する経験マイニング[1],および(b)Web情報の信頼性分析の支援を目的として,与えられた命題の裏付け情報を検索する言論マップ生成[2],という2種類のWeb情報編集技術の開発を例に,大規模知識獲得によって言語処理技術がどのように深化するか,今後の知識獲得研究に何を期待するかを論じたい.

  116. 1X-8 日本語文章における直接照応および間接照応の統合的解析(意味・談話解析,学生セッション,人工知能と認知科学,情報処理学会創立50周年記念)

    井之上 直也, 飯田 龍, 乾 健太郎, 松本 裕治

    全国大会講演論文集 72 (2) "2-541"-"2-542" 2010年3月8日

  117. Web情報分析のための大規模Webページの収集・選択・検索

    赤峯 享, 加藤 義清, 河原 大輔, レオン 末松豊インティ, 新里 圭司, 乾 健太郎, 黒橋 禎夫, 木俵 豊

    言語処理学会第16回年次大会 238-241 2010年3月

  118. Web情報の俯瞰的把握のための主要・対比・対立文の抽出と集約

    河原 大輔, 乾 健太郎, 黒橋 禎夫

    言語処理学会第16回年次大会 134-137 2010年3月

  119. Web ページの情報発信構成の同定

    加藤 義清, 河原 大輔, 乾 健太郎, 黒橋 禎夫

    言語処理学会第16回年次大会 90-93 2010年3月

  120. Resolving Direct and Indirect Anaphora for Japanese Definite Noun Phrases

    INOUE Naoya, IIDA Ryu, INUI Kentaro, MATSUMOTO Yuji

    自然言語処理 17 (1) 221-246 2010年1月10日

    DOI: 10.5715/jnlp.17.1_221  

    ISSN:1340-7619

  121. 問題を考える

    乾健太郎

    自然言語処理 17 (1) 2010年1月

  122. 言語情報編集のための広義モダリティ解析に向けて 招待有り

    乾 健太郎, 松吉 俊

    Japio 2009 YEAR BOOK 128-133 2009年11月1日

    出版者・発行元:一般財団法人日本特許情報機構

  123. 日本語文章の事象に対する判断情報アノテーション

    江口 萌, 松吉 俊, 佐尾 ちとせ, 乾 健太郎, 松本 裕治

    研究報告自然言語処理(NL) 2009 (5) 1-8 2009年9月21日

    出版者・発行元:情報処理学会

    ISSN:0919-6072

    詳細を見る 詳細を閉じる

    情報抽出や含意認識などの応用において、個々の事象に対して、その述語と項構造を認識するだけではなく、書き手が表明している態度や真偽判断、価値判断などの情報も解析し、その解析結果に基づいて情報を整理することは重要である。本研究では、このような情報をモダリティ情報と呼ぶ。本論文では、事象のモダリティ情報を表す次の 7 つ組のタグの体系を提案する:態度表明者、時制、仮想、態度、真偽判断、価値判断、焦点。このタグ体系に基づいて構築中である、約 4 万事象を対象としたコーパスの構築過程と現状についても報告する。Many NLP tasks including information extraction and textual entailment recognition involve the task of analyzing the modality status of each event mention in a given text on top of predicate-argument structure analysis. This paper reviews the literature of modality annotation and proposes a new annotation scheme to cover all the aspects of modality which are only partly treated in previous schemes. In our scheme, each event mention is annotated with seven slots: Source, Tense, Assumptional, Modality type, Authenticity, Sentiment, and Focus. The paper also reports on the present results of our manual annotation of a Japanese corpus consisting of about 40,000 event mentions, showing a reasonably high ratio of inter-annotator agreement.

  124. SIGVerse:社会的インタラクションを実装可能な知能ロボットシミュレータ

    稲邑哲也, 柴田智広, 瀬名秀明, 橋本敬, 川合伸幸, 宮下敬宏, 櫻井圭記, 清水正宏, 大武美保子, 細田耕, 梅田聡, 乾健太郎

    日本ロボット学会学術講演会予稿集(CD-ROM) 27th ROMBUNNO.3S2-05 2009年9月15日

  125. 複数文書から抽出した言明間の意味的関係の整理と関係付与(コーパス)

    村上浩司, 増田祥子, 松吉俊, NICHOLSEric, 乾健太郎, 松本裕治

    電子情報通信学会技術研究報告. NLC, 言語理解とコミュニケーション 109 (142) 31-36 2009年7月15日

    出版者・発行元:一般社団法人電子情報通信学会

    ISSN:0913-5685

    詳細を見る 詳細を閉じる

    我々は現在,Web情報の信憑性評価を行うために,あるトピックに関するWeb文書集合において,さまざまな視点や角度から述べられる言明を抽出し,それらの間の類似・対立・根拠等の関係を認識して,ユーザに言明と関係の情報を提示する言論マップ生成課題に取り組んでいる.本論文では言論マップ生成のために,Web上に存在する客観的表現や,意見などの主観的表現を扱うためには,どのような種類の意味的関係が必要であるかについて議論する.また,言明間意味的関係コーパスを構築するために我々が必要とする意味的関係を持つような用例対をWeb文書から効率よく収集する方法についても議論する.そして,言明間意味的関係コーパスを構築するためのアノテーション結果について報告し,これまでにアノテーションが終了している約1,500言明ペアについての評価に対する考察を行う.この言明間意味的関係コーパスは,今夏に試用版を公開する予定である.

  126. Annotating Statement Pairs Extracted from Web Documents with Semantic Relations

    MURAKAMI Koji, MASUDA Shouko, MASUDA Shouko, MATSUYOSHI Suguru, NICHOLS Eric, INUI Kentaro, MATSUMOTO Yuji

    電子情報通信学会技術研究報告 109 (142(NLC2009 1-6)) 31-36 2009年7月15日

    出版者・発行元:一般社団法人電子情報通信学会

    DOI: 10.1527/tjsai.22.227  

    ISSN:0913-5685

    詳細を見る 詳細を閉じる

    我々は現在,Web情報の信憑性評価を行うために,あるトピックに関するWeb文書集合において,さまざまな視点や角度から述べられる言明を抽出し,それらの間の類似・対立・根拠等の関係を認識して,ユーザに言明と関係の情報を提示する言論マップ生成課題に取り組んでいる.本論文では言論マップ生成のために,Web上に存在する客観的表現や,意見などの主観的表現を扱うためには,どのような種類の意味的関係が必要であるかについて議論する.また,言明間意味的関係コーパスを構築するために我々が必要とする意味的関係を持つような用例対をWeb文書から効率よく収集する方法についても議論する.そして,言明間意味的関係コーパスを構築するためのアノテーション結果について報告し,これまでにアノテーションが終了している約1,500言明ペアについての評価に対する考察を行う.この言明間意味的関係コーパスは,今夏に試用版を公開する予定である.

  127. 隠れ変数を持つ条件付き確率場による依存構造木の評価極性分類

    中川 哲治, 乾 健太郎, 黒橋 禎夫

    第192回自然言語処理研究会 2009年7月

  128. Webページの大規模収集・検索基盤

    赤峯 享, 加藤 義清, 河原 大輔, 新里 圭司, 乾 健太郎, 黒橋 禎夫, 木俵 豊

    情報処理学会研究会 Vol.2009-DBS-148 No.14 2009年7月

  129. 雑談対話のための評価表現を利用する相槌

    清水 友裕, 乾 健太郎, 松本 裕治

    言語・音声理解と対話処理研究会 55 7-12 2009年3月13日

    出版者・発行元:人工知能学会

    ISSN:0918-5682

  130. 事例の重み付けに基づく自動獲得されたコーパスの効果的な利用法と評価極性分類への応用

    中川 哲治, 乾 健太郎, 黒橋 禎夫

    電子情報通信学会技術研究報告. NLC, 言語理解とコミュニケーション 108 (408) 25-30 2009年1月19日

    出版者・発行元:一般社団法人電子情報通信学会

    ISSN:0913-5685

    詳細を見る 詳細を閉じる

    本稿では,自動獲得されたラベル付きデータを利用して,統計的な分類器に基づく評価極性分類の精度を改善させる方法について述べる.正解ラベルの付与された訓練データは機械学習に基づく自然言語処理を行う上で非常に重要であり,これまでにラベル付きデータを自動的に収集する方法がいくつか提案されている.しかしながらそのようなデータは誤りを含んでいる可能性があるため,訓練データとして直接利用するには適さない可能性がある.そこでこの問題に対処するために,ラベルの確信度により事例の重み付けを行う方法を提案する.実験の結果,提案手法を用いて自動獲得されたラベル付きデータを利用することにより,評価極性分類の精度を向上させることができた.

  131. SIGVerse:身体的認知・運動を伴う社会エージェントのためのシミュレータープラットフォーム

    稲邑哲也, 柴田智広, 瀬名秀明, 橋本敬, 川合伸幸, 宮下敬宏, 櫻井圭記, 清水正宏, 大武美保子, 細田耕, 梅田聡, 乾健太郎

    人工知能学会全国大会論文集(CD-ROM) 23rd (0) ROMBUNNO.2I2-1-2I21 2009年

    出版者・発行元:一般社団法人 人工知能学会

    ISSN:1347-9881

    詳細を見る 詳細を閉じる

    &lt;p&gt;従来までに,社会エージェントシステムを用いて経済や言語進化などのシミュレーションが実現されてきている.一方で,ロボティクスでは物理法則に基づく運動や,視聴覚等のセンシングのシミュレーションに基づいて知能行動の研究が行われている.本稿ではこの双方のシミュレーションを同時に実行可能なシステムを構築し,人間とロボットエージェント間の身体性認知・運動を考慮した社会的知能の研究プラットフォームを提案する&lt;/p&gt;

  132. Dependency Language Modeling Using KNN and PLSI

    Hiram Calvo, Kentaro Inui, Yuji Matsumoto

    MICAI 2009: ADVANCES IN ARTIFICIAL INTELLIGENCE, PROCEEDINGS 5845 136-+ 2009年

    出版者・発行元:SPRINGER-VERLAG BERLIN

    DOI: 10.1007/978-3-642-05258-3_12  

    ISSN:0302-9743

    詳細を見る 詳細を閉じる

    In this paper we present a comparison of two language models based on dependency triples. We explore using the verb only for predicting the most plausible argument as in selectional preferences, as well as using both the verb and argument for predicting another argument. This latter causes a problem of data sparseness that must be solved by different techniques for data smoothing. Based on our results on the K-Nearest Neighbor model (KNN) algorithm we conclude that adding more information is useful for attaining higher precision, while the PLSI model was inconveniently sensitive to this information, yielding better results for the simpler model (using the verb only). Our results suggest that combining the strengths of both algorithms would provide best results.

  133. Learning Co-relations of Plausible Verb Arguments with a WSM and a Distributional Thesaurus

    Hiram Calvo, Kentaro Inui, Yuji Matsumoto

    PROGRESS IN PATTERN RECOGNITION, IMAGE ANALYSIS, COMPUTER VISION, AND APPLICATIONS, PROCEEDINGS 5856 363-+ 2009年

    出版者・発行元:SPRINGER-VERLAG BERLIN

    DOI: 10.1007/978-3-642-10268-4_43  

    ISSN:0302-9743

    詳細を見る 詳細を閉じる

    We propose a model based on the Word Space Model for calculating the plausibility of candidate arguments given one verb and one argument. The resulting information can be used in co-reference resolution, zero-pronoun resolution or syntactic ambiguity tasks. Previous work such as Selectional Preferences or Semantic Frames acquisition focuses on this task using supervised resources, or predicting arguments independently from each other. On this work we explore the extraction of plausible arguments considering their co-relation, and using no more information than that provided by the dependency parser. This creates a data sparseness problem alleviated by using a distributional thesaurus built from the same data for smoothing. We compare our model with the traditional PLSI method.

  134. 結束性と首尾一貫性から見たゼロ照応解析(解析・抽出)

    飯田龍, 乾健太郎, 松本裕治

    情報処理学会研究報告. 自然言語処理研究会報告 2008 (90) 45-52 2008年9月17日

    出版者・発行元:一般社団法人情報処理学会

    ISSN:0919-6072

    詳細を見る 詳細を閉じる

    ゼロ照応解析の問題を結束性と首尾一貫性の観点から検討する.結束性の観点からは,Walkerのキャッシュモデルの実現方法を検討し,統計的機械学習に基づく実装を提案する.このキャッシュモデルを用いて文間ゼロ照応の先行詞候補削減を試み,評価実験を通じて先行詞同定時に解析対象とする先行詞候補を激減できたことを報告する.また,首尾一貫性の観点からは,含意関係認識で利用される推論知識獲得の手法を照応解析の手がかりとすることで解析精度にどのように影響するかについて調査する.新聞記事を対象に先行詞同定の実験を行い,導入した推論規則が解析に有効に働くことを示す.

  135. 事象間関係知識の整備と類似・対立認識への応用(単語・事象・オントロジー)

    松吉俊, 村上浩司, 増田祥子, 松本裕治, 乾健太郎

    情報処理学会研究報告. 自然言語処理研究会報告 2008 (90) 15-22 2008年9月17日

    出版者・発行元:一般社団法人情報処理学会

    ISSN:0919-6072

    詳細を見る 詳細を閉じる

    あるトピックについて書かれた大量のウェブ文書から、そのトピックについての深い理解を得るためには、それらの文書に記述されている個々の言論の信憑性や有効性を適切に判断し、かつ、暗に存在しているかもしれない言論間の対立関係を発見しようとすることが必要不可欠である。このような作業の支援を目的として、我々は、言論間に存在する類義や対立、根拠などの論理的関係を解析し、それらを言論マップと呼ぶ俯瞰図で明示するための技術の開発に取り組んでいる。本論文では、言論マップの生成に必要な事象間関係知識の整備に関して、既存の2つの言語資源の統合と、下位事象を内包する動詞項構造の記述について報告する。予備実験の結果、整備した事象間関係知識データベースを用いると、約8割の高い精度で言論間の類義関係を認識できることが分かった。

  136. Webページの著者の同定

    加藤 義清, 河原 大輔, 乾 健太郎, 黒橋 禎夫, 柴田 知秀

    第7回情報科学技術フォーラム(FIT2008) 2008年9月

  137. E-001 Web文書からの主観的・客観的評価表現の抽出(自然言語・音声・音楽,一般論文)

    中川 哲治, 川田 拓也, 乾 健太郎, 黒橋 禎夫

    情報科学技術フォーラム講演論文集 7 (2) 137-138 2008年8月20日

    出版者・発行元:FIT(電子情報通信学会・情報処理学会)運営委員会

  138. D-036 Web情報信頼性分析のためのページ外観情報の抽出(データベース,一般論文)

    赤峯 享, 加藤 義清, 乾 健太郎, 黒橋 禎夫

    情報科学技術フォーラム講演論文集 7 (2) 127-128 2008年8月20日

    出版者・発行元:FIT(電子情報通信学会・情報処理学会)運営委員会

  139. E-053 言論マップ生成のための事象間類似・対立関係の認識(自然言語・音声・音楽,一般論文)

    村上 浩司, 松吉 俊, 増田 祥子, 松本 裕治, 乾 健太郎

    情報科学技術フォーラム講演論文集 7 (2) 261-262 2008年8月20日

    出版者・発行元:FIT(電子情報通信学会・情報処理学会)運営委員会

  140. E-040 含意・矛盾認識のための事象間関係知識の整備(自然言語・音声・音楽,一般論文)

    松吉 俊, 村上 浩司, 松本 裕治, 乾 健太郎

    情報科学技術フォーラム講演論文集 7 (2) 231-232 2008年8月20日

    出版者・発行元:FIT(電子情報通信学会・情報処理学会)運営委員会

  141. 言論マップ生成課題 : 言説間の類似・対立の構造を捉えるために

    村上 浩司, 松吉 俊, 隅田 飛鳥, 森田 啓, 佐尾 ちとせ, 増田 祥子, 松本 裕治, 乾 健太郎

    情報処理学会研究報告. NL,自然言語処理研究会報告 186 55-60 2008年7月17日

  142. 2種類のアンカー情報と共起パターンの組み合わせによる事態間関係獲得

    阿部 修也, 乾 健太郎, 松本 裕治

    情報処理学会研究報告. NL,自然言語処理研究会報告 186 19-24 2008年7月17日

  143. 言論マップ生成課題 : 言説間の類似・対立の構造を捉えるために(情報分析・要約(テーマセッション1))

    村上浩司, 松吉俊, 隅田飛鳥, 森田啓, 佐尾ちとせ, 増田祥子, 松本裕治, 乾健太郎

    電子情報通信学会技術研究報告. NLC, 言語理解とコミュニケーション 108 (141) 55-60 2008年7月10日

    詳細を見る 詳細を閉じる

    Web文書には様々な情報が存在し、あるトピックについて多角的な言論などが述べられていることが多い。こうしたトピックに関わる種々の言論は、単純なクエリ検索だけでは広く網羅することができず、偏在する特定の立場の言論を中心的に捉えてしまう危惧がある.本稿ではWeb情報中の数的に優勢な立場の言論だけでなく、与えられたトピックに関して存在する多様な言論を抽出し、それらの言論間の類似、対立、含意等の論理的関係を解析してマップ化する言論マップ生成課題について論じる。また、述語項構造レベルの言論間の関係解析について、既存の事態関係知識を利用した予備実験について報告し、言論マップ生成のために必要な個々の技術課題について述べる。

  144. 2種類のアンカー情報と共起パターンの組み合わせによる事態間関係獲得(語彙知識)

    阿部修也, 乾健太郎, 松本裕治

    電子情報通信学会技術研究報告. NLC, 言語理解とコミュニケーション 108 (141) 19-24 2008年7月10日

    出版者・発行元:一般社団法人電子情報通信学会

    ISSN:0913-5685

    詳細を見る 詳細を閉じる

    行為-効果,行為-目的のような事態間の関係を大規模コーパスから自動的に獲得する.文内共起パターンを利用する手法では,事態間でどの項が共有されるかの知識を獲得することが難しい.そこで事態間で共有される名詞(アンカー)を用いて項共有情報を獲得し,文内共起パターンによる事態間関係と組み合わせることで項を必要とする事態間関係を獲得する.このとき2種類の異なるアンカーを用いることで,精度を保ったまま再現率を向上できることを確認した.

  145. 言論マップ生成課題 : 言説間の類似・対立の構造を捉えるために(情報分析・要約(テーマセッション1))

    村上浩司, 松吉俊, 隅田飛鳥, 森田啓, 佐尾ちとせ, 増田祥子, 松本裕治, 乾健太郎

    情報処理学会研究報告. 自然言語処理研究会報告 2008 (67) 55-60 2008年7月10日

    詳細を見る 詳細を閉じる

    Web文書には様々な情報が存在し、あるトピックについて多角的な言論などが述べられていることが多い。こうしたトピックに関わる種々の言論は、単純なクエリ検索だけでは広く網羅することができず、偏在する特定の立場の言論を中心的に捉えてしまう危惧がある.本稿ではWeb情報中の数的に優勢な立場の言論だけでなく、与えられたトピックに関して存在する多様な言論を抽出し、それらの言論間の類似、対立、含意等の論理的関係を解析してマップ化する言論マップ生成課題について論じる。また、述語項構造レベルの言論間の関係解析について、既存の事態関係知識を利用した予備実験について報告し、言論マップ生成のために必要な個々の技術課題について述べる。

  146. 2種類のアンカー情報と共起パターンの組み合わせによる事態間関係獲得(語彙知識)

    阿部修也, 乾健太郎, 松本裕治

    情報処理学会研究報告. 自然言語処理研究会報告 2008 (67) 19-24 2008年7月10日

    詳細を見る 詳細を閉じる

    行為-効果,行為-目的のような事態間の関係を大規模コーパスから自動的に獲得する.文内共起パターンを利用する手法では,事態間でどの項が共有されるかの知識を獲得することが難しい.そこで事態間で共有される名詞(アンカー)を用いて項共有情報を獲得し,文内共起パターンによる事態間関係と組み合わせることで項を必要とする事態間関係を獲得する.このとき2種類の異なるアンカーを用いることで,精度を保ったまま再現率を向上できることを確認した.

  147. 主要・対立表現の俯瞰的把握 - ウェブの情報信頼性分析に向けて

    河原 大輔, 黒橋 禎夫, 乾 健太郎

    情報処理学会 第186回自然言語処理研究会 2008年7月

  148. 4J-3 経験マイニングのための事実性解析(情報爆発時代におけるデータマイニング,一般セッション,「情報爆発」時代に向けた新しいIT基盤技術)

    原一夫, 東山昌彦, 乾健太郎, 松本裕治

    全国大会講演論文集 70 (5) "5-43"-"5-44" 2008年3月13日

  149. Web情報の信頼性検証のための情報分析システムWISDOM

    赤峯 亨, 宮森 恒, 加藤 義清, 中川 哲治, 乾 健太郎, 黒橋 禎夫, 木俵 豊

    言語処理学会 第14回年次大会 721-724 2008年3月

  150. Webページの情報発信者の同定とその関係の抽出

    加藤 義清, 乾 健太郎, 黒橋 禎夫

    言語処理学会 第14回年次大会 737-740 2008年3月

  151. Web上の客観的記述からの評価情報抽出に関する技術的検討

    中川 哲治, 宮森 恒, 赤峯 亨, 乾 健太郎, 黒橋 禎夫

    言語処理学会 第14回年次大会 344-347 2008年3月

  152. Webテキストにおける評価情報の整理・分類およびタグ付きコーパスの構築

    川田 拓也, 中川 哲治, 森井 律子, 宮森 恒, 赤峯 亨, 乾 健太郎, 黒橋 禎夫, 木俵 豊

    言語処理学会 第14回年次大会 524-527 2008年3月

  153. WWWからの知識獲得--自然言語処理における新展開 (特集 WWWを対象にした日本語研究)

    乾 健太郎, 鳥澤 健太郎

    日本語学 27 (2) 48-61 2008年2月

    出版者・発行元:明治書院

    ISSN:0288-0822

  154. 事態抽出のための事実性解析(Web情報アクセス)

    原一夫, 乾健太郎

    情報処理学会研究報告. 自然言語処理研究会報告 2008 (4) 75-80 2008年1月21日

    出版者・発行元:一般社団法人情報処理学会

    詳細を見る 詳細を閉じる

    本研究では事態の事実性解析に焦点をあてる。テキストに記述されるすべての事態を対象にした情報抽出を行う上で、事実性解析は述語項構造解析とともに必須の技術といえるが、これまでのところ、十分には研究されていない。本稿では、事実性解析の応用例としてブログ記事からの経験抽出を想定し、そのための事実性解析を時間情報、極性、話者態度の観点から抽象化する枠組みを提案する。また、ブログ記事を対象とする予備実験についても報告する。

  155. 事態抽出のための事実性解析(Web情報アクセス)

    原一夫, 乾健太郎

    情報処理学会研究報告. 情報学基礎研究会報告 2008 (4) 75-80 2008年1月21日

    出版者・発行元:一般社団法人情報処理学会

    ISSN:0919-6072

    詳細を見る 詳細を閉じる

    本研究では事態の事実性解析に焦点をあてる。テキストに記述されるすべての事態を対象にした情報抽出を行う上で、事実性解析は述語項構造解析とともに必須の技術といえるが、これまでのところ、十分には研究されていない。本稿では、事実性解析の応用例としてブログ記事からの経験抽出を想定し、そのための事実性解析を時間情報、極性、話者態度の観点から抽象化する枠組みを提案する。また、ブログ記事を対象とする予備実験についても報告する。

  156. 名詞化された事態表現への意味的注釈付け

    飯田龍, 小町守, 乾健太郎, 松本裕治

    言語処理学会年次大会発表論文集 14th 2008年

  157. 汎用アノテーションツールSLAT

    野口正樹, 三好健太, 徳永健伸, 飯田龍, 小町守, 乾健太郎

    言語処理学会年次大会発表論文集 14th 2008年

  158. テキスト情報の事実性解析

    森田啓, 佐尾 ちとせ, 松吉 俊, 松本 裕治, 乾 健太郎

    第7回情報科学技術フォーラム, 2008 259-260 2008年

    出版者・発行元:FIT(電子情報通信学会・情報処理学会)運営委員会

  159. Untitled

    Yoshikiyo Kato, Sadao Kurohashi, Kentaro Inui

    INTERNET RESEARCH 18 (2) 140-141 2008年

    出版者・発行元:EMERALD GROUP PUBLISHING LIMITED

    ISSN:1066-2243

  160. Information Credibility Analysis of Web Contents 査読有り

    Sadao Kurohashi, Susumu Akamine, Daisuke Kawahara, Yoshikiyo Kato, Tetsuji Nakagawa, Kentaro Inui, Yutaka Kidawara

    PROCEEDINGS OF THE SECOND INTERNATIONAL SYMPOSIUM ON UNIVERSAL COMMUNICATION 146-+ 2008年

    出版者・発行元:IEEE COMPUTER SOC

    DOI: 10.1109/ISUC.2008.79  

    詳細を見る 詳細を閉じる

    As computers and computer networks become more sophisticated, a vast amount of information and knowledge has been accumulated and circulated on the Web. They provide people with options regarding their daily lives and are starting to have a strong influence on governmental policies and business management. However, a crucial problem is that information on the Web is not necessarily credible. It is actually very difficult for human beings to judge information credibility and even more difficult for computers. However, computers can be used to develop a system that collects, organizes and relativises information and helps human beings view information from several viewpoints and judge information credibility. This paper introduces the information credibility criteria project at the National Institute of Information and Communications Technology, which aims to develop such a system, called WISDOM.

  161. Grasping major statements and their contradictions toward information credibility analysis of web contents 査読有り

    Daisuke Kawahara, Sadao Kurohashi, Kentaro Inui

    Proceedings - 2008 IEEE/WIC/ACM International Conference on Web Intelligence, WI 2008 393-397 2008年

    DOI: 10.1109/WIIAT.2008.289  

    詳細を見る 詳細を閉じる

    The World Wide Web contains wide variety of news reports, arguments, opinions, etc. that vary widely in quality. People judge the credibility of information on the Web for decision making in daily life. At present, while the quantity of information on the Web is explosively increasing, it is necessary to develop a system that supports such judgments. We have been developing an information credibility analysis system, WISDOM that considers the viewpoints of information contents, information senders, and information appearances. In this paper, as a viewpoint of information contents, we propose a method for providing a bird's eye view of major statements on a given topic and their contradictions. We evaluate the obtained statements in our experiments, and confirm the effectiveness of our approach. Furthermore, we discuss our future objectives. © 2008 IEEE.

  162. 情報の信頼性分析に向けた評価データおよびプロトタイプシステムWISDOM

    宮森恒, 赤峯享, 加藤義清, 兼岩憲, 角薫, 乾健太郎, 黒橋禎夫

    情報処理学会研究報告. 自然言語処理研究会報告 2007 (76) 103-108 2007年7月24日

    出版者・発行元:一般社団法人情報処理学会

    ISSN:0919-6072

    詳細を見る 詳細を閉じる

    本稿では,情報の信頼性を自然言語処理に基づいて分析する際に必要となる評価用データおよびプロトタイプシステムWISDOMについて述べる.われわれは,ウェブ上のテキストを主な対象として,情報信頼性を分析することを目指したプロジェクトを2006年4月より進めている.本プロジェクトでは,ウェブ上の情報の信頼性を,情報内容,情報発信者,情報外観,社会的評価といった4つの基準で捉えることを提案しており,これらを述語項構造を単位とする自然言語処理によって論理的に分析・組織化することを目指している.本稿で述べる評価用データは,これら種々の分析処理の学習・検証用データとして構築されたものであり,時事問題,医療問題等の20トピックを選定し,各100ウェブページを収集して,各評価尺度のデータを人手で付与したものである.また,情報信頼性を多角的に評価するプロトタイプシステムWISDOMを開発した.本システムを用いて上記評価尺度で条件を様々に変化させて情報閲覧することにより,興味のトピックについて,信頼できる情報をより確実に見極めることができるようになる.

  163. 言語表現された推論知識による帰結文生成の技術的課題

    大西良明, 乾健太郎, 松本裕治

    情報処理学会研究報告. 自然言語処理研究会報告 2007 (76) 55-60 2007年7月24日

    出版者・発行元:一般社団法人情報処理学会

    ISSN:0919-6072

    詳細を見る 詳細を閉じる

    計算機で高度な言語処理を行うためには,言語解析だけでなく世界知識に基づいた高度な推論が必要となる.本稿では,自然言語で表現された推論規則,より具体的には述語項構造で表現された事態対によって記述された推論規則を使って推論を行う場合の技術的課題について論じる.実際に,国語辞典の語釈文から推論規則を抽出し,これを入力文に適用する予備的実験を行った.誤り分析の結果, (1)岩波国語辞典の語義の粒度で述語の多義性を解消したとしても,さらに粒度の細かい多義性によって推論誤りが生じる場合が少なくないこと, (2)言語モデルによる帰結文のフィルタリングによって推論誤りをある程度回避できること, (3)推論規則のカバレッジを確保するには格交替を柔軟に扱える機構が必要なこと, (4)項に「AのB」の構造が内在する場合に「B」の顕在化・潜在化を柔軟に扱う仕組みが必要なこと,などが明らかになった.

  164. 情報の信頼性分析に向けた評価データおよびプロトタイプシステムWISDOM(QA・Web検索)

    宮森恒, 赤峯享, 加藤義清, 兼岩憲, 角薫, 乾健太郎, 黒橋禎夫

    電子情報通信学会技術研究報告. NLC, 言語理解とコミュニケーション 107 (158) 103-108 2007年7月17日

    出版者・発行元:一般社団法人電子情報通信学会

    ISSN:0913-5685

    詳細を見る 詳細を閉じる

    本稿では,情報の信頼性を自然言語処理に基づいて分析する際に必要となる評価用データおよびプロトタイプシステムWISDOMについて述べる.われわれは,ウェブ上のテキストを主な対象として,情報信頼性を分析することを目指したプロジェクトを2006年4月より進めている.本プロジェクトでは,ウェブ上の情報の信頼性を,情報内容,情報発信者,情報外観,社会的評価といった4つの基準で捉えることを提案しており,これらを述語項構造を単位とする自然言語処理によって論理的に分析・組織化することを目指している.本稿で述べる評価用データは,これら種々の分析処理の学習・検証用データとして構築されたものであり,時事問題,医療問題等の20トピックを選定し,各100ウェブページを収集して,各評価尺度のデータを人手で付与したものである.また,情報信頼性を多角的に評価するプロトタイプシステムWISDOMを開発した.本システムを用いて上記評価尺度で条件を様々に変化させて情報閲覧することにより,興味のトピックについて,信頼できる情報をより確実に見極めることができるようになる.

  165. 言語表現された推論知識による帰結文生成の技術的課題(語彙1)

    大西良明, 乾健太郎, 松本裕治

    電子情報通信学会技術研究報告. NLC, 言語理解とコミュニケーション 107 (158) 55-60 2007年7月17日

    詳細を見る 詳細を閉じる

    計算機で高度な言語処理を行うためには,言語解析だけでなく世界知識に基づいた高度な推論が必要となる.本稿では,自然言語で表現された推論規則,より具体的には述語項構造で表現された事態対によって記述された推論規則を使って推論を行う場合の技術的課題について論じる.実際に,国語辞典の語釈文から推論規則を抽出し,これを入力文に適用する予備的実験を行った.誤り分析の結果,(1)岩波国語辞典の語義の粒度で述語の多義性を解消したとしても,さらに粒度の細かい多義性によって推論誤りが生じる場合が少なくないこと,(2)言語モデルによる帰結文のフィルタリングによって推論誤りをある程度回避できること,(3)推論規則のカバレッジを確保するには格交替を柔軟に扱える機構が必要なこと,(4)項に「AのB」の構造が内在する場合に「B」の顕在化・潜在化を柔軟に扱う仕組みが必要なこと,などが明らかになった.

  166. Classifying Information Sender of Web Documents

    Yoshikiyo Kato, Sadao Kurohashi, Kentaro Inui

    In Proceedings of the First International Symposium on Universal Communication 238-241 2007年6月

  167. Evaluation Data and Prototype System WISDOM for Information Credibility Analysis

    Hisashi Miyamori, Susumu Akamine, Yoshikiyo Kato, Ken Kaneiwa, Kaoru Sumi, Kentaro Inui, Sadao Kurohashi

    In Proceedings of the First International Symposium on Universal Communication 234-237 2007年6月

  168. Web文書の情報発信者クラス分類

    加藤 義清, 黒橋 禎夫, 乾 健太郎

    言語処理学会 第13回年次大会 891-894 2007年3月

  169. NAISTテキストコーパス : 述語項構造と共参照関係のアノテーション(解析・対話)

    飯田龍, 小町守, 乾健太郎, 松本裕治

    情報処理学会研究報告. 自然言語処理研究会報告 2007 (7) 71-78 2007年1月26日

    出版者・発行元:一般社団法人情報処理学会

    ISSN:0919-6072

    詳細を見る 詳細を閉じる

    本稿では,日本語書き言葉を対象とした述語項構造と共参照のタグ付与について議論する.述語項構造や共参照解析は形態素・構文解析などの基盤技術と自然言語処理の応用分野とを繋ぐ重要な技術であり,これらの問題の主要な解析手法はタグ付与コーパスに基づく学習ベースの手法である.この手法で利用するための大規模な訓練データが必要となるが,これまでに日本語を対象にした大規模なタグ付きコーパスは存在しなかった.また,既存のコーパス作成に関する研究で採用されているタグ付与の基準は,言語の違いや我々が対象としたい解析と異なるために,そのまま採用することができない.そこで,既存のいくつかのタグ付与の仕様を比較し,我々のタグ付与作業で採用する基準について吟味する.また,実際に京都コーパス第3.0版の文章を対象にタグ付与の仕様について検討した結果とタグ付与の際に問題となった点や今後検討すべき点について報告する.

  170. 含意関係計算のための事態オントロジーの開発に向けて(言語理解とオントロジーシンポジウム)

    乾健太郎, 竹内孔一, 藤田篤

    電子情報通信学会技術研究報告. NLC, 言語理解とコミュニケーション 106 (518) 13-18 2007年1月24日

    出版者・発行元:一般社団法人電子情報通信学会

    ISSN:0913-5685

  171. 日本語書き言葉を対象とした述語項構造と共参照関係のアノテーション:NAISTテキストコーパス開発の経験から

    飯田龍, 小町守, 乾健太郎, 松本裕治

    言語処理学会年次大会発表論文集 13th 2007年

  172. 事態性名詞の項構造解析における共起尺度と構文パターンの有効性の分析

    小町守, 飯田龍, 乾健太郎, 松本裕治

    言語処理学会年次大会発表論文集 13th 2007年

  173. Opinion Mining from Web Documents: Extraction and Structurization

    KOBAYASHI Nozomi, INUI Kentaro, MATSUMOTO Yuji

    Inf Media Techol (Web) 2 (1) 326-337 (J-STAGE) 2007年

    DOI: 10.11185/imt.2.326  

    ISSN:1881-0896

  174. 事態含意名詞を用いた事態間関係知識の獲得(語彙・概念の獲得と同義語)

    阿部修也, 乾健太郎, 松本裕治

    情報処理学会研究報告. 自然言語処理研究会報告 2006 (124) 95-100 2006年11月22日

    出版者・発行元:一般社団法人情報処理学会

    ISSN:0919-6072

    詳細を見る 詳細を閉じる

    「〜するために〜する」のような2つの事態表現の共起パターンを利用してテキストコーパスから事態間の関係の知識を獲得する事態間関係知識獲得の問題について論じる.知識獲得の精度と規模を両立するには,意味的な制限の強い特殊な共起パターンを数多く用意することが望ましい.この問題に対し,本稿では,(a)事態を直接的あるいは間接的に指す事態含意名詞を利用すること,(b)実体間関係抽出のための共起パターン獲得技術を事態間関係獲得に応用することを検討し,これまでに得られた実験結果を報告する.

  175. 事態含意名詞を用いた事態間関係知識の獲得

    阿部 修也, 乾 健太郎, 松本 裕治

    情報処理学会研究報告. NL,自然言語処理研究会報告 176 95-100 2006年11月22日

  176. 言語情報処理における辞書と語彙概念構造 (まえがき)

    加藤 恒昭, 乾 健太郎, 竹内 孔一

    レキシコンフォーラム (2) 1-25 2006年4月

    出版者・発行元:ひつじ書房

  177. 言い換え計算モデルの工学的実現への語彙意味論的アプローチ (まえがき)

    乾 健太郎, 藤田 篤

    レキシコンフォーラム (2) 27-55 2006年4月

    出版者・発行元:ひつじ書房

  178. 語彙概念構造に基づく日本語動詞の統語・意味特性の記述 (まえがき)

    竹内 孔一, 乾 健太郎, 藤田 篤

    レキシコンフォーラム (2) 85-120 2006年4月

    出版者・発行元:ひつじ書房

  179. 意見情報の抽出/構造化のタスク仕様に関する考察

    小林のぞみ, 乾健太郎, 松本裕治

    情報処理学会研究報告. 自然言語処理研究会報告 2006 (1) 111-118 2006年1月12日

    出版者・発行元:一般社団法人情報処理学会

    ISSN:0919-6072

    詳細を見る 詳細を閉じる

    ネット上に分散する個人の意見を検索・分類・要約する意見マイニングを目的として, その部分問題である意見情報抽出の課題設定のあり方を論じる.〈評価対象, 属性, 評価値〉からなる従来の構造化の枠組みを拡張し, 新しい仕様に基づいて意見タグ付きコーパスを試作することにより, どの範囲の情報を意見として抽出するのが現実的か, 抽出した意見をどのような構造で表現すべきかを考察した.

  180. 共起用例と名詞の出現パターンを用いた動作性名詞の項構造解析

    小町守, 飯田龍, 乾健太郎, 松本裕治

    言語処理学会年次大会発表論文集 12th 2006年

  181. ゼロ照応解析のための統語的パターンの学習

    飯田龍, 乾 健太郎, 松本 裕治

    第5回情報科学技術フォーラム,FIT, 2006 81-84 2006年

    出版者・発行元:FIT(電子情報通信学会・情報処理学会)運営委員会

  182. Opinion mining as extraction of attribute-value relations

    Nozomi Kobayashi, Ryu Iida, Kentaro Inui, Yuji Matsumoto

    NEW FRONTIERS IN ARTIFICIAL INTELLIGENCE 4012 470-481 2006年

    出版者・発行元:SPRINGER-VERLAG BERLIN

    DOI: 10.1007/11780496_49  

    ISSN:0302-9743

    詳細を見る 詳細を閉じる

    This paper addresses the task of extracting opinions from a given document collection. Assuming that an opinion can be represented as a tuple (Subject, Attribute, Value), we propose a computational method to extract such tuples from texts. In this method, the main task is decomposed into (a) the process of extracting Attribute-Value pairs from a given text and (b) the process of judging whether an extracted pair expresses an opinion of the author. We apply machine-learning techniques to both subtasks. We also report on the results of our experiments and discuss future directions.

  183. 形態・構文的パターンを用いた言い換えコーパスの構築(抽出, 言い換え)

    藤田篤, 乾健太郎

    情報処理学会研究報告. 自然言語処理研究会報告 2005 (117) 143-150 2005年11月21日

    出版者・発行元:一般社団法人情報処理学会

    ISSN:0919-6072

    詳細を見る 詳細を閉じる

    語彙・構文的言い換えの中には, 形態・構文的パターンに基づいて一括りにできるものの, 表現を構成する語の統語・意味的な特性に依存して言い換えの可否や言い換え方が決まる現象が少なくない.たとえば, 複合語を構成語に分解するような言い換え, 機能動詞構文の言い換え, 態や格の交替, 種々の動詞交替, 語彙的派生などはこの語彙構成的言い換えの範疇に含まれる.我々は現在, これら語彙構成的言い換えに関わる語の統語・意味的な特性を明らかにするため, および言い換え生成技術の定量的評価のために, 個々の言い換えクラスごとに言い換え事例集(言い換えコーパス)を構築している.本稿では, 言い換え前後の表現の形態・構文的パターンと既存の言い換え生成システムを用いて言い換え事例を半自動的に収集する手法について述べる.また, 日本語の機能動詞構文の言い換え, 動詞の自他交替を対象とした予備試行の結果を報告する.

  184. 照応性判定を含む名詞句照応解析の実験と分析(抽出と解析)

    飯田龍, 乾健太郎, 松本裕治

    情報処理学会研究報告. 自然言語処理研究会報告 2005 (94) 93-100 2005年9月29日

    出版者・発行元:一般社団法人情報処理学会

    ISSN:0919-6072

    詳細を見る 詳細を閉じる

    本稿では, 我々が提案した照応解析手法[9]が照応性判定に必要な先行文脈の情報と局所文脈の情報を効果的に併用できていることについて説明する.さらに, 我々の照応解析モデルはSoonら[21]やNgら[20]の従来の学習に基づく照応解析モデルの利点をすべて継承しながらも, この既存手法の欠点を克服している点について議論する.提案手法を評価するために日本語名詞句照応解析の実験を行い, 従来の学習に基づく手法の性能を改善できたことを報告する.また, 提案手法を用いた解析の結果を人手で分析し, 今後の方向性を論じる.

  185. 分類の根拠を明示した動詞語彙概念構造辞書の構築(辞書と辞典)

    竹内孔一, 乾健太郎, 藤田篤, 竹内奈央, 阿部修也

    情報処理学会研究報告. 情報学基礎研究会報告 2005 (94) 123-130 2005年9月29日

    詳細を見る 詳細を閉じる

    本研究では語彙概念構造を利用した動詞の辞書データベースを構築する.語彙概念構造は構文上での語の振る舞いを基に意味構造を記述するもので, こうした辞書データは言語処理において特に言語生成, 例えば翻訳や言い換え, 要約といった応用に必須となる言語資源である.提案する辞書データは語彙概念構造のみを記述するのではなく, 語彙概念構造の分析で明らかにする語の振る舞い, ならびに分析の基となる統語テストを記述するため, 分類根拠がわかりやすく透明性の高い辞書データとなることが期待できる.

  186. 照応性判定を含む名詞句照応解析の実験と分析(抽出と解析)

    飯田龍, 乾健太郎, 松本裕治

    情報処理学会研究報告. 情報学基礎研究会報告 2005 (94) 93-100 2005年9月29日

    出版者・発行元:一般社団法人情報処理学会

    詳細を見る 詳細を閉じる

    本稿では, 我々が提案した照応解析手法[9]が照応性判定に必要な先行文脈の情報と局所文脈の情報を効果的に併用できていることについて説明する.さらに, 我々の照応解析モデルはSoonら[21]やNgら[20]の従来の学習に基づく照応解析モデルの利点をすべて継承しながらも, この既存手法の欠点を克服している点について議論する.提案手法を評価するために日本語名詞句照応解析の実験を行い, 従来の学習に基づく手法の性能を改善できたことを報告する.また, 提案手法を用いた解析の結果を人手で分析し, 今後の方向性を論じる.

  187. Webサイトへのアクセシビリティ向上を目的とした難語の平易化(高齢者支援/肢体不自由者支援/一般)

    中野智子, 遠藤淳, 菅原昌平, 乾健太郎, 藤田篤

    電子情報通信学会技術研究報告. WIT, 福祉情報工学 105 (186) 11-14 2005年7月7日

    出版者・発行元:一般社団法人電子情報通信学会

    ISSN:0913-5685

    詳細を見る 詳細を閉じる

    Webページに含まれるテキスト情報は, 利用者の知識や閲覧環境, 学習経歴などによって分かりやすさが異なる.特に, 高齢者や障害者などの情報弱者にとってこの問題は深刻である.本稿では, 問題の一つである&quot;難語&quot;を取り上げ, 言い換え技術を用いてWebページ上の難語の自動変換する実験について報告する.対象ドメインを限定すれば, 現在の資源・技術のチューニングによって言い換えの正確性と網羅性をある程度確保できる見通しを得た.

  188. 英国で見つけた歴史の試験問題に思う

    乾健太郎

    人工知能学会誌 20 (2) 243-244 2005年3月1日

    ISSN:0912-8085

  189. 意見抽出を目的とした機械学習による属性-評価値対同定(属性抽出)

    飯田龍, 小林のぞみ, 乾健太郎, 松本裕治, 立石健二, 福島俊一

    情報処理学会研究報告. 自然言語処理研究会報告 2005 (1) 21-28 2005年1月11日

    出版者・発行元:一般社団法人情報処理学会

    ISSN:0919-6072

    詳細を見る 詳細を閉じる

    本稿では, 文章に記述されている意見を抽出するタスクについて述べる.我々は, 意見を&lt;対象, 属性, 評価値&gt;の3つ組として定義し, 文章からその3つ組を抽出する手法を提案する.具体的には, 意見抽出の問題を, (a)辞書に存在する属性候補集合と評価値候補集合から評価値候補と対となる属性を同定する問題と, (b)同定した対が意見性を持つか否かを判定する問題の2つの問題に分解し, それぞれ機械学習に基づく手法を用いて解析することにより, 属性-評価値対を同定する.提案手法を用いて評価実験を行った結果を報告するとともに, 今後の展望についても議論する.

  190. Anaphora resolution by antecedent identification followed by anaphoricity determination

    Ryu Iida, Kentaro Inui, Yuji Matsumoto

    ACM Transactions on Asian Language Information Processing 4 (4) 417-434 2005年

    DOI: 10.1145/1113308.1113312  

    ISSN:1530-0226

    詳細を見る 詳細を閉じる

    We propose a machine learning-based approach to noun-phrase anaphora resolution that combines the advantages of previous learning-based models while overcoming their drawbacks. Our anaphora resolution process reverses the order of the steps in the classification-then-search model proposed by Ng and Cardie [2002b], inheriting all the advantages of that model. We conducted experiments on resolving noun-phrase anaphora in Japanese. The results show that with the selection-then-classification-based modifications, our proposed model outperforms earlier learning-based approaches. © 2005 ACM.

  191. Exploiting lexical conceptual structure for paraphrase generation

    A Fujita, K Inui, Y Matsumoto

    NATURAL LANGUAGE PROCESSING - IJCNLP 2005, PROCEEDINGS 3651 908-919 2005年

    出版者・発行元:SPRINGER-VERLAG BERLIN

    DOI: 10.1007/11562214_79  

    ISSN:0302-9743

    詳細を見る 詳細を閉じる

    Lexical Conceptual Structure (LCS) represents verbs as semantic structures with a limited number of semantic predicates. This paper attempts to exploit how LCS can be used to explain the regularities underlying lexical and syntactic paraphrases, such as-verb alternation, compound word decomposition, and lexical derivation. We propose a paraphrase generation model which transforms LCSs of verbs, and then conduct an empirical experiment taking the paraphrasing of Japanese light-verb constructions as an example. Experimental results justify that syntactic and semantic properties of verbs encoded in LCS are useful to semantically constrain the syntactic transformation in paraphrase generation.

  192. テキストから属性関係を抽出する(表理解,アライメント,抽出)

    高橋哲朗, 乾健太郎, 松本裕治

    情報処理学会研究報告. 自然言語処理研究会報告 2004 (108) 19-24 2004年11月4日

    出版者・発行元:一般社団法人情報処理学会

    ISSN:0919-6072

    詳細を見る 詳細を閉じる

    質問応答で用いる知識をオフラインで獲得するために(対象物,属性名,属性値)という三つ組の属性関係をテキスト中から抽出するタスクを考え,抽象化したパタンによりドメインを限定せずに三つ組の候補を抽出し,統計量を使ってそれらをフィルタリングする手法を提案した.実験の結果,この手法を用いることによりパタンのみで抽出した場合に比べて高い精度で属性関係の三つ組を抽出できることを示せた.本研究ではまた対象物と属性値だけをパタンにより抽出し,それらの間の関係の推定を行なった.その結果,既知の属性については高い精度で属性名を推定できることを明らかにした.

  193. テキストから属性関係を抽出する(表理解,アライメント,抽出)

    高橋哲朗, 乾健太郎, 松本裕治

    電子情報通信学会技術研究報告. NLC, 言語理解とコミュニケーション 104 (416) 19-24 2004年10月28日

    出版者・発行元:一般社団法人電子情報通信学会

    ISSN:0913-5685

    詳細を見る 詳細を閉じる

    質問応答で用いる知識をオフラインで獲得するために(対象物,属性名,属性値)という三つ組の属性関係をテキスト中から抽出するタスクを考え,抽象化したパタンによりドメインを限定せずに三つ組の候補を抽出し,統計量を使ってそれらをフィルタリングする手法を提案した.実験の結果,この手法を用いることによりパタンのみで抽出した場合に比べて高い精度で属性関係の三つ組を抽出できることを示せた.本研究ではまた対象物と属性値だけをパタンにより抽出し,それらの間の関係の推定を行なった.その結果,既知の属性については高い精度で属性名を推定できることを明らかにした.

  194. Web文書集合からの意見情報抽出と着眼点に基づく要約生成(Webマイニング)(テーマ:「Webマイニングによる情報活動と自然言語処理」その他一般)

    立石健二, 福島俊一, 小林のぞみ, 高橋哲朗, 藤田篤, 乾健太郎, 松本裕治

    情報処理学会研究報告. 自然言語処理研究会報告 2004 (93) 1-8 2004年9月16日

    出版者・発行元:一般社団法人情報処理学会

    詳細を見る 詳細を閉じる

    本稿では、Web文書から意見を抽出し、それらをレーダーチャートの形式で要約/視覚化する意見抽出分類システムを提案する。Webの意見は、商品購入の際の情報収集、市場調査等のマーケティング、企業のリスク管理等、さまざまな目的での利用が考えられる。Webの意見の収集/分析に関する研究には2つの課題がある、対象とするWeb文書から意見に該当する箇所を抽出すること、抽出した意見を要約/視覚化することである。本システムは、この2つの課題を3つ組{対象物,属性,評価}のモデルと情報抽出の手法を用いて解決する。本システムを車に関するレビューサイトの100記事を対象として評価したところ抽出精度が適合率82%,再現率52%であり、システムが出力したレーダーチャートと人手で作成したレーダーチャートが類似することを確認した。

  195. Web文書集合からの意見情報抽出と着眼点に基づく要約生成(Webマイニング)(テーマ:「Webマイニングによる情報活用と自然言語処理」その他一般)

    立石健二, 福島俊一, 小林のぞみ, 高橋哲朗, 藤田篤, 乾健太郎, 松本裕治

    情報処理学会研究報告. 情報学基礎研究会報告 2004 (93) 1-8 2004年9月16日

    詳細を見る 詳細を閉じる

    本稿では、Web文書から意見を抽出し、それらをレーダーチャートの形式で要約/視覚化する意見抽出分類システムを提案する。Webの意見は、商品購入の際の情報収集、市場調査等のマーケティング、企業のリスク管理等、さまざまな目的での利用が考えられる。Webの意見の収集/分析に関する研究には2つの課題がある、対象とするWeb文書から意見に該当する箇所を抽出すること、抽出した意見を要約/視覚化することである。本システムは、この2つの課題を3つ組{対象物,属性,評価}のモデルと情報抽出の手法を用いて解決する。本システムを車に関するレビューサイトの100記事を対象として評価したところ抽出精度が適合率82%,再現率52%であり、システムが出力したレーダーチャートと人手で作成したレーダーチャートが類似することを確認した。

  196. 情報検索対話におけるユーザ発話の検索要求単位への分割(対話インタフェース)(第5回音声言語シンポジウム)

    徳久良子, 寺嶌立太, 脇田敏裕, 乾健太郎

    情報処理学会研究報告. SLP, 音声言語情報処理 2003 (124) 247-252 2003年12月18日

    出版者・発行元:一般社団法人情報処理学会

    ISSN:0919-6072

    詳細を見る 詳細を閉じる

    従来の情報検索対話システムの多くは,システム応答のタイミングが適切でないためにシステムがユーザの検索要求を正確に理解できないという問題があった.そこで我々は,ユーザからシステムにイニシアチブが移行するタイミングに注目し,検索要求単位タグつきコーパスを作成した.作成したコーパスに基づいて機械学習によりユーザ発話を検索要求単位にチャンキングしたところ,形態素情報のみを用いた場合でも81.1%が正しく判定できることが分かった.また,音声対話システムヘの応用に向けて音声認識精度と検索要求単位判定精度との関係を調べたところ,70%以上の音声認識率があれば認識誤りが全くない場合とほぼ同等の精度で検索要求単位を判定できることが分かった.

  197. 情報検索対話におけるユーザ発話の検索要求単位への分割(第5回音声言語シンポジウム : 対話インタフェース)

    徳久良子, 寺嶌立太, 脇田敏裕, 乾健太郎

    電子情報通信学会技術研究報告. SP, 音声 103 (520) 1-6 2003年12月12日

    詳細を見る 詳細を閉じる

    従来の情報検索対話システムの多くは,システム応答のタイミングが適切でないためにシステムがユーザの検索要求を正確に理解できないという問題があった.そこで我々は,ユーザからシステムにイニシアチブが移行するタイミングに注目し,検索要求単位タグつきコーパスを作成した.作成したコーパスに基づいて機械学習によりユーザ発話を検索要求単位にチャンキングしたところ,形態素情報のみを用いた場合でも81.1%が正しく判定できることが分かった.また,音声対話システムへの応用に向けて音声認識精度と検索要求単位判定精度との関係を調べたところ,70%以上の音声認識率があれば認識誤りが全くない場合とほぼ同等の精度で検索要求単位を判定できることが分かった.

  198. 情報検索対話におけるユーザ発話の検索要求単位への分割(第5回音声言語シンポジウム : 対話インタフェース)

    徳久良子, 寺嶌立太, 脇田敏裕, 乾健太郎

    電子情報通信学会技術研究報告. NLC, 言語理解とコミュニケーション 103 (518) 1-6 2003年12月12日

    出版者・発行元:一般社団法人電子情報通信学会

    ISSN:0913-5685

    詳細を見る 詳細を閉じる

    従来の情報検索対話システムの多くは,システム応答のタイミングが適切でないためにシステムがユーザの検索要求を正確に理解できないという問題があった.そこで我々は,ユーザからシステムにイニシアチブが移行するタイミングに注目し,検索要求単位タグつきコーパスを作成した.作成したコーパスに基づいて機械学習によりユーザ発話を検索要求単位にチャンキングしたところ,形態素情報のみを用いた場合でも81.1%が正しく判定できることが分かった.また,音声対話システムへの応用に向けて音声認識精度と検索要求単位判定精度との関係を調べたところ,70%以上の音声認識率があれば認識誤りが全くない場合とほぼ同等の精度で検索要求単位を判定できることが分かった.

  199. 情報検索対話におけるユーザ発話の検索要求単位への分割

    徳久 良子, 寺嶌 立太, 脇田 敏裕, 乾 健太郎

    電子情報通信学会技術研究報告. NLC, 言語理解とコミュニケーション 103 (518) 1-6 2003年12月12日

  200. Effects of structural matching and paraphrasing in question answering

    T Takahashi, K Nawata, K Inui, Y Matsumoto

    IEICE TRANSACTIONS ON INFORMATION AND SYSTEMS E86D (9) 1677-1685 2003年9月

    出版者・発行元:IEICE-INST ELECTRONICS INFORMATION COMMUNICATIONS ENG

    ISSN:0916-8532

    詳細を見る 詳細を閉じる

    In this paper, we propose an answer seeking algorithm for question answering that integrates structural matching and paraphrasing, and report the results of our empirical evaluation conducted with the aim of examining effects of incorporating those two components. According to the results, the contribution of structural matching and paraphrasing was not so large as expected. Based on error analysis, we conclude that structural matching-based approaches to answer seeking require technologies for (a) coreference resolution, (b) processing of parse forests instead of parse trees, and (c) large-scale acquisition of paraphrase patterns.

  201. 自動生成した言い換え文における動詞結合価誤りの自動検出手法(言語知識・解析・言い換え)

    藤田篤, 乾健太郎, 松本裕治

    情報処理学会研究報告. 自然言語処理研究会報告 2003 (76) 53-60 2003年7月25日

    出版者・発行元:一般社団法人情報処理学会

    ISSN:0919-6072

    詳細を見る 詳細を閉じる

    本稿では,語彙・構文的言い換えにおいて頻繁に生じる動詞結合価誤りの検出方法を提案する.われわれは,コーパスから獲得した大規模な正例に基づいて結合価の適格さを定量化するモデルと,人手で収集した小規模の負例に基づいて結合価の不適格さを定量化するモデルを構築し,これら2つのアンサンブルによって,精度の高い誤り検出器を実現した.また,能動学習の採用によって,誤り検出に村して貢献度が高い負例を効率良く収集できることを確認した.

  202. 機械学習によるゼロ代名詞同定の一方法

    飯田龍, 乾健太郎, 高村大也, 松本裕治

    情報処理学会研究報告. 自然言語処理研究会報告 2003 (23) 161-168 2003年3月6日

    出版者・発行元:一般社団法人情報処理学会

    ISSN:0919-6072

    詳細を見る 詳細を閉じる

    センタリング理論のような言語学的な知見を,機械学習を用いた照応解析に統合する一方法を提案する.先行研究であるSoonら(2001)の解析手法に対して,我々は(i)局所的な要素を考慮した素性(センタリング素性)の追加と,(ii)先行詞候補間を比較するモデル(トーナメントモデル)の2点を改良した,この提案手法を用いて日本語ゼロ代名詞の同定を行い,先行研究の手法より精度よく先行詞の同定ができたことを報告する.

  203. テキストマイニングによる評価現象の収集

    小林のぞみ, 乾健太郎, 松本裕治, 立石健二, 福島俊一

    情報処理学会研究報告. 自然言語処理研究会報告 2003 (23) 77-84 2003年3月6日

    出版者・発行元:一般社団法人情報処理学会

    ISSN:0919-6072

    詳細を見る 詳細を閉じる

    近年,Web上に多数存在する掲示板などの文書から,意見情報を自動的に収集・解析する技術への関心が高まっている.このような意見情報の抽出には,評価を表す表現が重要な手がかりとなるが,評価表現には「コクがある」「液晶がきれい」といった領域依存の表現が多数存在するため,人手で書き尽くすことは困難である.そこで,我々は,テキストマイニングの技術を応用し,評価対象表現,属性表現,評価表現の共起パタンを利用して,これら領域依存表現を効率的に収集することを試みた.本稿では,共起パタンに基づく属性・評価表現の半自動的収集方法を提案し,「コンピュータ」と「ビール」の2つの領域を対象に行った経験的評価を報告する.

  204. 半構造化文書に対する情報探索のための対話履歴管理 (テーマ:一般)

    山下 亜希子, 乾 健太郎, 松本 裕治

    言語・音声理解と対話処理研究会 36 45-50 2002年11月7日

    出版者・発行元:人工知能学会

    ISSN:0918-5682

  205. 機械学習による点字表記の解析 : 点訳支援システム構築に向けて

    菅野亜紀, 高岡裕, 米田隆一, 乾健太郎, 松本裕治

    電子情報通信学会技術研究報告. TL, 思考と言語 102 (254) 9-14 2002年7月22日

    出版者・発行元:一般社団法人電子情報通信学会

    ISSN:0913-5685

    詳細を見る 詳細を閉じる

    機械点訳精度向上を目的として、点訳処理の際に問題点となる点字表記の多様性の解析と機械学習による分かち書きの検討を行ったので報告する。点字には表記規則が定められているが、分かち書き規則に存在する自由度(=曖昧性)が、表記の多様性を引き起こしている。点字表記の多様性を解明することは点訳支援システム構築に向けて意義深いが、これまでの研究は表記規則による多様性の予測に限定される傾向があった。しかし、Support Vector Machineに基づく機械学習では、点字文書に存在する分かち書きの多様性を解析可能であり、結果として分かち書きの精度向上も期待される。

  206. 接続助詞「ため」を含む複文から因果関係知識を獲得する

    乾孝司, 乾健太郎, 松本裕治

    情報処理学会研究報告. 自然言語処理研究会報告 2002 (66) 171-178 2002年7月15日

    出版者・発行元:一般社団法人情報処理学会

    ISSN:0919-6072

    詳細を見る 詳細を閉じる

    大規模な電子化テキストから因果関係知識を自動的に獲得する方法について論じる.我々の目標は,事態間の因果関係が明示的に表現されている修辞的表現から,それを一貫性のある表現として成り立たせている(広義の)因果関係に関する背景知識を同定し,蓄積することである.まず,接続助詞「ため」「のに」「たら」を含む複文を分析し,7つの基本概念を設定した.次に,接続助詞「ため」を含む複文を対象として,生節,従属節の間に内在する関係を規則に基づいて自動判別する実験を行った.その結果,再現率は低いが,effectやcauseなどの関係概念では比較的高い適合率を得た.また,同一の知識源からより多くの知識を獲得する方法について検討した.

  207. テキストの構文的類似度の評価方法について

    高橋哲朗, 乾健太郎, 松本裕治

    情報処理学会研究報告. 自然言語処理研究会報告 2002 (66) 163-170 2002年7月15日

    出版者・発行元:一般社団法人情報処理学会

    ISSN:0919-6072

    詳細を見る 詳細を閉じる

    与えられた2つの構文木の類似度を効率的に評価する方法について論じる.Collinsが提案したTree Kernelの拡張し,構文木間の一般的な類似度評価方法を与える.この評価方法は,大きく3種類の類似性尺度を提供し,計算量のオーダをO(|T_1||T_2|)に抑えながら,対応ノード間の類似度の定量化やノードの飛び越えを許す照合を実現することができる.また,提案した構文的類似度評価方法が質問応答タスクに応用できる可能性についても議論する.我々の評価方法は,従来の手法に比べて計算コストのオーバヘッドを定数倍に抑えたまま,より大域的な構造の照合をより柔軟に行うことができる.

  208. 接続助詞「ため」を含む複文から因果関係知識を獲得する

    乾孝司, 乾健太郎, 松本裕治

    電子情報通信学会技術研究報告. NLC, 言語理解とコミュニケーション 102 (200) 73-80 2002年7月9日

    出版者・発行元:一般社団法人電子情報通信学会

    ISSN:0913-5685

    詳細を見る 詳細を閉じる

    大規模な電子化テキストから因果関係知識を自動的に獲得する方法について論じる.我々の目標は,事態間の因果関係が明示的に表現されている修辞的表現から,それを一貰性のある表現として成り立たせている(広義の)因果関係に関する背景知識を同定し,蓄積することである.まず,接続助詞「ため」「のに」「たら」を含む複文を分析し,7つの基本概念を設定した.次に,接続助詞「ため」を含む複文を対象として,主節,従属節の間に内在する関係を規則に基づいて自動判別する実験を行った.その結果,再現率は低いが,effectやcauseなどの関係概念では比較的高い適合率を得た.また,同一の知識源からより多くの知識を獲得する方法について検討した.

  209. テキストの構文的類似度の評価方法について

    高橋哲朗, 乾健太郎, 松本裕治

    電子情報通信学会技術研究報告. NLC, 言語理解とコミュニケーション 102 (200) 65-72 2002年7月9日

    出版者・発行元:一般社団法人電子情報通信学会

    ISSN:0913-5685

    詳細を見る 詳細を閉じる

    与えられた2つの構文木の類似度を効率的に評価する方法について論じる.Collinsが提案したTree Kernelの拡張し,構文木間の一般的な類似度評価方法を与える.この評価方法は,大きく3種類の類似性尺度を提供し,計算量のオーダをO(&amp;mid;T_1&amp;mid;&amp;mid;T_2&amp;mid;)に抑えながら,対応ノード間の類似度の定量化やノードの飛び越えを許す照合を実現することができる.また,提案した構文的類似度評価方法が質問応答タスクに応用できる可能性についても議論する.我々の評価方法は,従来の手法に比べて計算コストのオーバヘッドを定数倍に抑えたまま,より大域的な構造の照合をより柔軟に行うことができる.

  210. 語彙的言い換えに必要な知識の部品化

    藤田篤, 乾健太郎

    情報処理学会研究報告. 自然言語処理研究会報告 2002 (44) 31-38 2002年5月23日

    出版者・発行元:一般社団法人情報処理学会

    ISSN:0919-6072

    詳細を見る 詳細を閉じる

    我々はこれまでに,語彙・構文的言い換えの事例研究を通じて,言い換えに必要な言語知識の種類を明らかにするとともに,それらを実装し,各種語彙・構文的言い換えを実現してきた.本稿では,先行研究ならびに我々のグループの研究で得られた知見を横並びに分析し,さまざまな言い換えに用いることができる部品として再構成することを提案する.具体的には,言い換え後の不適格性を修正する知識に焦点を当てる.まず,先行研究で報告されている言い換え規則と事例分析に基づいて部品を定義した.また,さまざまな語彙・構文的言い換えにおいて,各部品の再利用性を調査した.

  211. 語釈文を利用した「p/n 辞書」の作成 (テーマ なぜインタラクションでなければならないか?)

    小林 のぞみ, 乾 孝司, 乾 健太郎

    言語・音声理解と対話処理研究会 33 45-50 2001年11月5日

    出版者・発行元:人工知能学会

    ISSN:0918-5682

  212. Kura:統一的かつ宣言的知識記述に基づく言い換えエンジン

    岩倉, 友哉, 高橋, 哲朗, 飯田, 龍, 乾, 健太郎

    第63回全国大会講演論文集 2001 (1) 277-278 2001年9月26日

  213. Kura : 統一的かつ宣言的記述法に基づく言い換え知識の開発環境

    高橋 哲朗, 岩倉 友哉, 飯田 龍, 乾 健太郎

    電子情報通信学会技術研究報告. TL, 思考と言語 101 (243) 23-30 2001年7月24日

    出版者・発行元:一般社団法人電子情報通信学会

    ISSN:0913-5685

    詳細を見る 詳細を閉じる

    本稿では以下の3つの特徴を持つ言い換えエンジンKURAについて, そのアーキテクチャと実装に焦点を当てて報告する.(a)言い換え処理を, 構造変換と棄却・修正とに分けて行う.(b)構造変換規則および棄却・修正のための規則を統一的で宣言的に記述できる.(c)大規模な言い換え実験に耐えるように高速化の工夫が施されている.またシステムを用いて実験を行うための環境も構築したので, その動作例もあわせて報告する.

  214. 日本語文生成における照応表現の選択

    橋本 さち恵, 乾 健太郎, 白井 清昭, 徳永 健伸, 田中 穂積

    情報処理学会研究報告自然言語処理(NL) 2001 (54) 33-40 2001年5月31日

    出版者・発行元:一般社団法人情報処理学会

    ISSN:0919-6072

    詳細を見る 詳細を閉じる

    指示詞や代名詞,あるいはゼロ代名詞といった様々な照応表現の中から適切なものを選択することは,自然な文章を生成するための重要な課題である.本研究は,日本語を対象とし,日本語文生成の際に適切な照応表現を選択するモデルを構築することを目的とする.まず,実際のテキストに出現した照応表現を分析し,どのような情報が適切な照応表現を選択するための手がかりとなるのかを調査した.次に,分析した結果を元に,照応表現を選択するモデルを人手で構築した.作成したモデルの有効性を評価する実験を行った結果,約96?%の事例について,適切な照応表現を選択できることがわかった.As there are many kinds of anaphoric expressions such as demonstrative, pronoun, zero anaphor etc., selecting a natural anaphoric expression is one of the important task in the natural language generation. In this paper, we aimed at constructing a new model to select an appropriate anaphoric expression in the generation of Japanese. First, we analyzed the text and explored effective features for selecting anaphoric expressions. Then, we manually constructed the model according to the result of this analysis. Experimental results indicated that 96\% anaphoric expression produced by our model is appropriate, showing the validness of it.

  215. 読解支援を目的とするテキスト簡単化の実現に向けて : 課題と方法論

    乾 健太郎

    電子情報通信学会技術研究報告. TL, 思考と言語 101 (61) 51-58 2001年5月11日

    出版者・発行元:一般社団法人電子情報通信学会

    ISSN:0913-5685

    詳細を見る 詳細を閉じる

    我々は, テキスト簡単化によって先天的聾者の文章読解を支援するソフトウェア技術の研究開発プロジェクトを進めている. 「テキスト簡単化」とは, 一般のテキストに構文的, 語彙的言い換え操作を適用することによって, これをより平易で理解しやすいテキストに変換する作業である. 本稿では, テキスト簡単化を構成する2つのサブタスク, (a) 言い換え生成と (b) テキスト評価に問題を分けて整理し, それぞれに対する我々のアプローチとこれまでの成果, および今後の展望について報告する.

  216. 言語コーパスにおける感情生起要因と感情クラスの注釈づけ (テーマ:コーパスを利用した談話・対話研究)

    徳久 良子, 乾 健太郎, 徳久 雅人

    言語・音声理解と対話処理研究会 31 9-14 2001年3月2日

    出版者・発行元:人工知能学会

    ISSN:0918-5682

  217. 規模とコストを考慮した感情タグつき言語コーパスの作成方法

    徳久良子, 乾健太郎, 徳久雅人, 岡田直之

    電子情報通信学会総合大会講演論文集, シンポジウム:心を持つロボット,対話するロボット 514-515 2001年3月

    出版者・発行元:一般社団法人電子情報通信学会

  218. 連用節主節化に関する規則の追試と洗練

    神田 慎哉, 藤田 篤, 乾 健太郎

    人工知能学会全国大会論文集 15 1-4 2001年

    出版者・発行元:人工知能学会

    ISSN:1347-9881

  219. 感情の生起とその反応

    乾健太郎, 徳久雅人, 徳久良子, 岡田直之

    特集:感情のモデルと工学的応用の動向, 日本ファジィ学会誌 12 (6) 741-751 2000年12月

    DOI: 10.3156/jfuzzy.12.6_23  

  220. 名詞言い換えコーパスの作成環境

    藤田 篤, 乾 健太郎, 乾 裕子

    電子情報通信学会技術研究報告. TL, 思考と言語 100 (480) 53-60 2000年11月27日

    出版者・発行元:一般社団法人電子情報通信学会

    ISSN:0913-5685

    詳細を見る 詳細を閉じる

    名詞や動詞などの内容語の言い換えでは, 言い換えの生成に必要な大規模な知識をどのように獲得するかが当面の課題になる.知識のソースには国語辞典の語釈文やシソーラスの同概念語が考えられるが, 言い換えの対象語を語釈文や同概念語と置き換えるだけのナイーブな方法では構文・意味的制約を満たす言い換えは実現できない.これに対し, 単語間の共起制約や語釈文と文脈の文法的整合性といった自明な制約を実装し, ある程度良質の言い換え事例を半自動的かつ大規模に生成する環境を構築した.また, この環境を用いて生成された事例を分析し, 語彙的言い換えの技術的な問題点を整理した.

  221. ここまできた自然言語処理-例文の収集とその利用-:例文を使って文の解析をしよう

    乾 健太郎, 白井 清昭

    情報処理 41 (7) 763-768 2000年7月15日

    出版者・発行元:一般社団法人情報処理学会

    ISSN:0447-8053

    詳細を見る 詳細を閉じる

    構文解析は自然言語処理における重要な要素技術の1つであるが,1つの入力文に対して,誤りを含む非常に多くの構文構造の候補が生成されることが問題となる.これを解決するために,たとえば選択制約など,人手で作成した規則によって誤りとなる候補を除去する手法が研究されてきたが,このような手法は網羅性,システムの保守性などに問題がある.

  222. 聾者向け文章読解支援のための文可読性基準の調査

    山本 聡美, 乾 健太郎, 野上 優, 藤田 篤, 乾 裕子

    情報処理学会研究報告自然言語処理(NL) 2000 (11) 127-134 2000年1月27日

    出版者・発行元:一般社団法人情報処理学会

    ISSN:0919-6072

    詳細を見る 詳細を閉じる

    我々は,テキスト簡単化によって先天的聾者の文章読解を支援するソフトウェア技術の研究開発プロジェクトを進めている.「テキスト簡単化」とは,一般のテキストに構文的・語棄的言い換え操作を適用することによって,これをより平易で理解しやすいテキストに変換する作業である.本稿ではまず,プロジェクトの概要を紹介し,主要な研究課題を整理する.つぎに,テキスト簡単化において重要な資源となる可読性基準と言い換えパタンの構築に向けて,高等聾学校教諭を対象に可読性に関するアンケート調査を行ったので,その概要および分析手順,分析結果を報告する.This paper describes our ongoing research project on text simplification for congenitally deaf people. Text simplification we are aiming at is the task of offering a deaf reader a syntactic and lexical paraphrase of a given text for assisting her/him to understand what it means. In this paper, we particularly focus on our questionnaire survey on the readability of Japanese sentences for deaf people, presenting an overview and the results we have so far obtained.

  223. 統計的部分係り受け解析における係り受け確率の利用法 -コーパス中の構文タグ誤りの検出-

    乾 孝司, 乾 健太郎

    情報処理学会研究報告自然言語処理(NL) 1999 (95) 15-22 1999年11月25日

    出版者・発行元:一般社団法人情報処理学会

    ISSN:0919-6072

    詳細を見る 詳細を閉じる

    品質の高いコーパスを作成するためには,構文解析器によって自動的にタグづけした後,それを人手で修正する作業が不可欠である.コーパス中のタグ誤りを効率的に見つける方法があれば,コーパス修正の人的コストを大幅に削減できると考えられる.本稿では,統計的部分係り受け解析方式によって係り受け確率を推定し,これを係り受けタグの誤り検出に利用する方法について論じる.京大コーパスを用いて実験をおこなった結果,係り受け確率がある程度誤り検出に利用できるという見通しが得られた.We have been exploring a way of enhancing the current state of the art of statistical parsing by reintroducing the notion of partial parsing, which we call probabilistic partial parsing. Among the various advantages of probabilistic partial parsing, in this paper, we discuss the feasibility of applying it to the task of error detection in treebanks. This task is to retrieve erroneous tags from the all tags that are inconsistent with the parser's outputs. The results of our preliminary experiments on the Kyoto Japanese corpus shows that the scheme of our probabilistic partial parsing improves the performance of this retrieval task.

  224. 相手の情緒を理解する対話システムの構築に向けて : 情緒タグつき対話コーパスの構築

    徳久 良子, 徳久 雅人, 乾 健太郎, 岡田 直之

    電子情報通信学会技術研究報告. TL, 思考と言語 99 (237) 13-20 1999年7月26日

    出版者・発行元:一般社団法人電子情報通信学会

    ISSN:0913-5685

    詳細を見る 詳細を閉じる

    対話中に変化する相手 (ユーザ) の情緒を推定し, システムの応答プランニングに利用する技術の研究開発は, ヒューマンフレンドリな対話システムを構築するための有効なアプローチの一つと考えられる. このようなアプローチについてはすでに萌芽的な研究がいくつか見られるものの, ユーザの情緒の推定に必要な機構や情緒推定と応答プランニングの相互作用についてはまだほとんど明らかになっていない. そこで我々は, これらの問題を解明する手段の一つとして, 情緒タグつき対話コーパスの構築を検討している. 本稿では, 情緒タグつき対話コーパスを構築することの現実性について, 我々がこれまでに行った予備調査の結果を報告し, その中で明らかになった問題点について議論する.

  225. 品詞タグつきコーパスにおける品詞体系の変換

    乾 健太郎, 脇川 浩和

    電子情報通信学会技術研究報告. NLC, 言語理解とコミュニケーション 99 (228) 31-38 1999年7月23日

    出版者・発行元:一般社団法人電子情報通信学会

    ISSN:0913-5685

    詳細を見る 詳細を閉じる

    近年, 信頼性の高い品詞・構文タグつきコーパスに対する需要の増大にともなって, コーパスを共有・再利用することの重要性がますます大きくなっている. しかし, 既存のタグつきコーパスでは基礎とする品詞体系が統一されておらず, そのことが共有・再利用の障害となっている. このような背景から本稿では, 既存のコーパスの品詞・構文タグを別の品詞体系に基づく品詞・構文タブに変換するアルゴリズムについて論じる. 本稿で提案する手法では, ターゲット側品詞体系に基づく文法・辞書でコーパスを形態素・構文解析することによって半自動的にタグ付けを行う. このとき生じる曖昧性は, ソース側タグ情報を最大限に利用することによって高い精度で解消することができる.

  226. 品詞タグつきコーパスにおける品詞体系の変換

    乾 健太郎, 脇川 浩和

    情報処理学会研究報告自然言語処理(NL) 1999 (62) 87-94 1999年7月22日

    出版者・発行元:一般社団法人情報処理学会

    ISSN:0919-6072

    詳細を見る 詳細を閉じる

    近年,信頼性の高い品詞・構文タグつきコーパスに対する需要の増大にともなって,コーパスを共有・再利用することの重要性がますます大きくなっている.しかし,既存のタグつきコーパスでは基礎とする品詞体系が統一されておらず,そのことが共有・再利用の障害となっている.このような背景から本稿では,既存のコーパスの品詞・構文タグを別の品詞体系に基づく品詞・構文タグに変換するアルゴリズムについて論じる.本稿で提案する手法では,ターゲット側品詞体系に基づく文法・辞書でコーパスを形態素・構文解析することによって半自動的にタグ付けを行う.このとき生じる曖昧性は,ソース側タグ情報を最大限に利用することによって高い精度で解消することができる.The problems in reusing the POS-tag information of an existing corpus are in the gap between different tag sets; corpora are annotated in terms of different tag sets. While the recent efforts for standardizing tags are important, we still need to explore techniques for the (semi-)automatic conversion between different tag sets in order to maximally reuse the existing tagged corpora. This paper presents an NLP-based method for the conversion between Japanese POS-tag sets, and reports the results of our preliminary experiment.

  227. Empirical Support for New Probabilistic Generalized LR Parsing

    Sornlertlamvanich Virach, Inui Kentaro, Tanaka Hozumi, TOKUNAGA Takenobu, TAKEZAWA Toshiyuki

    自然言語処理 = Journal of natural language processing 6 (3) 3-22 1999年4月10日

    出版者・発行元:言語処理学会

    ISSN:1340-7619

  228. 統計的構文解析における構文的統計情報と語彙的統計情報の統合について

    白井 清昭, 乾 健太郎, 徳永 健伸, 田中 穂積

    自然言語処理 = Journal of natural language processing 5 (3) 85-106 1998年7月10日

    出版者・発行元:言語処理学会

    ISSN:1340-7619

    詳細を見る 詳細を閉じる

    本論文では,構文解析の曖昧性解消を行うために,構文的な統計情報と語彙的な統計情報を統合する手法を提案する.我々が提案する統合的確率言語モデルは,構文的優先度などの構文的な統計情報を反映する構文モデルと,単語の出現頻度や単語の共起関係などの語彙的な統計情報を反映する語彙モデルの2つの下位モデルから成る.この統計的確率言語モデルは,構文的な統計情報と語彙的な統計情報を同時に学習する過去の多くのモデルと異なり,両者を個別に学習する点に特徴がある.構文的な統計情報と語彙的な統計情報を独立に取り扱うことにより,それぞれの統計情報を異なる言語資源から独立に学習することができるだけでなく,それぞれの統計情報が曖昧性解消においてどのような効果を果たすのかを容易に分析することができる.この統計的確率言語モデルを評価するために,日本語文の文節の係り受け解析を行った.構文モデルを用いたときの文節の正解率は73.38%となり,ベースラインに比べて11.70%向上した.また,構文モデルと語彙モデルを組み合わせることにより,文節の正解率はさらに10.96%向上し84.34%となった.この結果,本研究で提案する枠組において,語彙的な統計情報は構文的な統計情報と同程度に曖昧性解消に貢献することを確認した. : In this paper, we propose a new framework of statistical language modeling integrating syntactic statistics and lexical statistics. Our model consists of two submodels, the syntactic model and lexical model. The syntactic model reflects syntactic statistics, such as structural preferences, whereas the lexical model reflects lexical statistics, such as occurrence of each word and word collocations. One of the characteristics of our model is that it learns both types of statistics separately, although many previous models learn them simultaneously. Learning each submodel separately enables us to use a different language source for different submodels, and to make understanding of each submodel&#039;s behavior much easier. We conducted a preliminary experiment, where our model was applied to the disambiguation of dependency structures of Japanese sentences. The syntactic model achieved 73.38% in Bunsetu phrase accuracy, which is 11.70 points above the baseline, and when incorporating the lexical model with the syntactic model, further 10.96 point gain was achieved, to 84.34%. Thus the contribution of lexical statistics for disambiguation is as great as that of syntactic statistics in our framework.

  229. Probabilistic GLR Parsing: A New Formalization and Its Impact on Parsing Performance

    INUI Kentaro, SORNLERTLAMVANICH Virach, TANAKA Hozumi, TOKUNAGA Takenobu

    自然言語処理 5 (3) 33-52 1998年

    出版者・発行元:一般社団法人 言語処理学会

    ISSN:1340-7619

    詳細を見る 詳細を閉じる

    This paper presents a new formalization of probabilistic GLR (PGLR) language modeling for statistical parsing.Our model inherits its essential features from Briscoe and Carroll&#039;s generalized probabilistic LR model (Briscoe and Carroll 1993), which takes context of parse derivation into account by assigning a probability to each LR parsing action according to its left and right context. Briscoe and Carroll&#039;s model, however, has a drawback in that it is not formalized in any probabilistically well-founded way, which may degrade its parsing performance. Our formulation overcomes this drawback with a few significant refinements, while maintaining all the advantages of Briscoe and Carroll&#039;s modeling. In this paper, we discuss the formal and qualitative aspects of our PGLR model, illustrating the qualitative differences between Briscoe and Carroll&#039;s model and our model, and their expected impact on parsing performance.

  230. 統合的確率モデルを用いた日本語文解析

    白井 清昭, 乾 健太郎, 徳永 健伸, 田中 穂積

    情報処理学会研究報告自然言語処理(NL) 1997 (109) 43-50 1997年11月20日

    出版者・発行元:一般社団法人情報処理学会

    詳細を見る 詳細を閉じる

    我々は構文的優先度,語彙的従属関係を同時に取り扱う統合的確率言語モデルを提案している.このモデルの特長は構文的優先度と語彙的従属関係を互いに独立に取り扱う点にある.これにより,両者を独立に学習することができるだけでなく,両者がそれぞれ曖昧性解消にどれだけ有効に作用するのかを容易に評価できる.本稿では,このモデルを用いて日本語文の係り受け解析実験を行った結果について報告し,構文的優先度,語粟的従属関係のそれぞれが文節の係り先の正解率の向上に大きく貢献することを示す.また,解析に失敗した原因について調査を行い,その主な要因と本稿で提案するモデルにおける対処法について論ずる.We propose a new statistical language model which integrates lexical association statistics with syntactic preferences, while maintaining the modularity of these different statistics types, facilitating both the training of the model and analysis of its behavior. In this paper, we report the results of an empirical evaluation of our model, in which the model is applied to the disambiguation of Japanese sentence dependency structures. The results show that both syntactic preferences and lexical associations significantly raise the accuracy, which is the ratio of the number of Bunsetu phrases whose modifiee is correctly identified, to the total phrase number. We also discuss further room for improvement based on our error analysis.

  231. 意思決定理論に基づく発話プランニング

    乾 健太郎, 徳永 健伸, 田中 穂積, Kentaro Inui, Takenobu Tokunaga, Hozumi Tanaka, Dept. of Computer Science Tokyo Institute of Technology, Dept. of Computer Science Tokyo Institute of Technology, Dept. of Computer Science Tokyo Institute of Technology

    人工知能学会誌 = Journal of Japanese Society for Artificial Intelligence 12 (5) 760-769 1997年9月1日

    出版者・発行元:人工知能学会

    ISSN:0912-8085

    詳細を見る 詳細を閉じる

    In human-computer dialogue, a diverse range of situations arise from uncertainty in reasoning regarding the user's beliefs and plans. The dialogue system is required to plan the contents of its utterances under such uncertain conditions. This paper presents a decision-theoretic framework of content planning for dialogue systems. In our framework, the uncertain domain knowledge and user's model are both declaratively described as probabilistic constraints, which represent uncertain situations as probabilistic distributions over possible worlds. We estimate this probabilistic distribution using a Bayesian network. Given a utility function to evaluate the degree of goal attainment in each possible world, the system chooses the content of its utterance based on the expected utility maximization principle. Our framework can perform content planning under uncertain conditions without any procedural planning strategy.

  232. 単語多義性解消法の比較検討

    藤井 敦, 乾 健太郎, 徳永 健伸, 田中 穂積

    情報処理学会研究報告自然言語処理(NL) 1997 (53) 45-52 1997年5月26日

    出版者・発行元:一般社団法人情報処理学会

    詳細を見る 詳細を閉じる

    本論文はコーパスに基づく単語多義性解消法の比較実験について報告する。近年提案されている手法の多くは、与えられた入力と単語語義を教示した例文の間の類似度に基づいて多義性解消を行う。日本語動詞を対象とした多義性解消の実験において、異なる類似度計算法を比較した結果、(a)動詞と補語の統語構造、(b)人手で作成したシソーラスを導入することによって最大で30%以上の多義性解消精度の向上が得られた。This paper describes an extensive comparative evaluation of recent corpus-based word sense disambiguation techniques, focusing around ten Japanese verbs. The basis of this task is the computation of the similarity between a given input and examples which have been annotated with its verb sense, and we compare different methods modeling this process. Through our experiments and discussion, we found the effective components of word sense disambiguation to be: (a) the use of the syntactic relation between a target verb and its complements (case pattern), and (b) the use of existing thesauri to approximate human knowledge.

  233. Incorporating Probabilistic Parsing into an LR Parser : LR Table Engineering (4)

    SORNLERTLAMVANICH Virach, INUI Kentaro, SHIRAI Kiyoaki, TANAKA Hozumi, TOKUNAGA Takenobu, TAKEZAWA Toshiyuki

    情報処理学会研究報告. NL,自然言語処理研究会報告 119 (53) 61-68 1997年5月26日

    出版者・発行元:一般社団法人情報処理学会

    詳細を見る 詳細を閉じる

    We propose a probabilistic model for incorporating probability into a (G)LR parser. The model is formalized based on stack transition during parsing distinguishing it from the existing models proposed by Wright and Wrigley, and Briscoe and Carroll. Our model produces a remarkable improvement in statistical parsing with probability. Associating probabilities directly to actions in an LR parsing table, and theoretically requiring only one probability for each action promise model trainability and potential extension to other related tasks.

  234. 最大エントロピー法を用いた単語bigramの推定

    白井 清昭, 乾 健太郎, 徳永 健伸, 田中 穂積

    情報処理学会研究報告自然言語処理(NL) 1996 (114) 21-28 1996年11月18日

    出版者・発行元:一般社団法人情報処理学会

    詳細を見る 詳細を閉じる

    単語bigramを推定する際には,一般にデータがスパースであるので,何らかのスムージングが必要となる.本稿では,シソーラスを利用した単語の抽象化によるスムージングの手法を分類し,それらとスムージングの結果得られる単語bigramの品質との関係について考察した.そして,最大エントロピー法による確率モデルの推定方法が単語bigramの推定に有効であることに着目した.また,最大エントロピー法をそのまま単語bigramの推定に利用するには計算量が大きいという問題点を解決するために,これを高速化するいくつかの手法を提案した.推定した単語bigramを用いて複合名詞の解析実験を行い,提案した手法の有効性を確認した.To estimate lexical bigram distribution, smoothing methods are generally applied to overcome concerns of data sparseness. First, we discuss the relation between smoothing methods using a thesaurus and the quality of lexical bigram distribution estimated by them. We conclude that maximum entropy methods are suitable for estimating lexical bigram distribution. As the computational cost associated with maximum entropy methods is too large for lexical bigram distribution estimation, we propose several methods to suppress the overhead. We conducted an experiment to analyze Japanese compound nouns, and the result indicated that our method is better than existing smoothing methods.

  235. 種々の制約を統合した統計的日本語文解析

    乾 健太郎, 白井 清昭, 徳永 健伸, 田中 穂積

    情報処理学会研究報告. NL,自然言語処理研究会報告 116 (114) 35-42 1996年11月18日

    出版者・発行元:一般社団法人情報処理学会

    詳細を見る 詳細を閉じる

    本稿では, 形態素解析・構文解析・多義性解消からなる複合的問題に統計的手法を適用するために, 個別の問題に対する既存の解決法をどのように拡張し, 組み合わせればよいかについて論じ, 構文モデル・語彙モデル・語義モデルからなる統合的言語モデルを提案する. 提案するモデルは, 確率文法, 係り受け関係の距離の分布, 隣接する品詞の従属関係, 語彙的従属関係を反映する. モデルの主な特徴は, 語彙的従属関係を評価するための従属係数という量を導入することにより既存手法のいくつかに確率的解釈を与え, それをモデルに組み込んだ点, 格フレーム構造を導入することにより構文的依存関係と意味的依存関係が同形でない場合に対処できるようにした点である.

  236. 効用最大化法に基づく多義性解消用事例の選択的収集

    藤井 敦, 乾 健太郎, 徳永 健伸, 田中 穂積

    情報処理学会研究報告音声言語情報処理(SLP) 1996 (55) 91-97 1996年5月27日

    出版者・発行元:一般社団法人情報処理学会

    ISSN:0919-6072

    詳細を見る 詳細を閉じる

    本論文は、多義性解消システムに用いる事例ベースを効率的に収集する手法を提案する。従来、事例への語義付与にかかる人間の負担、及び事例ベースのサイズが問題とされてきた。本手法の特長は、人間がその事例の語義を教えてシステムを訓練することがそれ以降の事例収集に及ぼす影響、すなわち効用を定量化し、効用の最も高い事例を選択的に収集する点にある。さらに本論文は、1000文以上の例文を含むコーパスを用いた実験について報告する。ランダムに事例を抽出する手法との比較実験の結果、本手法を用いて収集した事例ベースは、より少ない人間の負担、より小さいサイズで、より高い多義性解消の精度を実現できることが確認された。This paper proposes an efficient example selection method for example-based word sense disambiguation systems. To construct a practical size database, a considerable overhead for manual sense disambiguation is required. Our method is characterized by the reliance on the notion of the utility of training: the degree to which each example is informative for future example selection when used for the training of the system. The system progressively collects examples by selecting those with great utility. The paper reports the effectivity of our method through an experiment on over one thousand sentences. According to the comparative experiment with random example selection, our method reduced the overhead without the degeneration of the performance of the system.

  237. 多義性解消に用いる事例の獲得

    藤井 敦, 乾 健太郎, 徳永 健伸, 田中 穂積

    言語処理学会年次大会発表論文集 = Proceedings of the ... annual meeting of the Association for Natural Language Processing 2 261-264 1996年3月26日

  238. 日本語学習者用電子化辞書の開発に向けて

    白井 太三, 乾 健太郎, 太田 武昭, 徳永 健伸, 田中 穂積

    言語処理学会年次大会発表論文集 = Proceedings of the ... annual meeting of the Association for Natural Language Processing 2 405-408 1996年3月26日

  239. 単一化に基づく漸進的発話生成 (第16回 SLUD)

    坂庭 克幸, 乾 健太郎, 徳永 健伸

    合同研究会AIシンポジウム (7) 45-50 1996年

    出版者・発行元:人工知能学会

  240. 対話における即応的/熟考的な談話プランニングの制御について

    杉山 聡, 乾 健太郎, 長尾 確, 徳永 健伸, 田中 穂積

    全国大会講演論文集 51 105-106 1995年9月20日

    詳細を見る 詳細を閉じる

    本稿では,対話者がタスクプランを構築したり遂行したりすることを目的として情報伝達を行なう対話(タスク指向対話)における即応性/熟考性の切り換えの制御について論じる.即応的な対話システムは浅い推論にもとづいて発話するので,応答時間は短い反面,対話相手が聞いたことしか答えないなどふるまいが不親切になりやすい.一方,熟考的なシステムは深い推論をするので,対話相手に対し協調的にふるまうが,応答に時間がかかる.対話システムは状況に依存して即応性/熟考性を適切に切り換えられることが望ましいが,そのためには熟考的に推論することの効用と時間コストのトレードオフの最適化が必要である.しかしながら,従来のタスク指向対話システムの多くは可能なかぎり熟考的に推論を行なうように設計されており,時間コストを十分に考慮していない.この問題に対し本稿では,「浅い推論→即応的発話→少し深い推論→少し熟考的な発話→・・・→非常に深い推論→非常に熟考的な発話」というように推論と発話の実行をインタリーブさせ,推論と発話の繰り返しの途中で随時対話相手の発話が割り込めるという枠組について述べる.これは推論と行為の実行をインタリーブさせるという意味で一種のリアクティブプランニング[2]である.この枠組では,対話相手が自分に必要十分な情報を受けとった時点でシステムの発話に割り込むと期待できるので,対話相手にとってどの程度熟考的な発話が必要かを対話相手自身に判断させることができる.これによって,過度に複雑な機構を使わなくても状況に依存して熟考性の程度を変化させることができると考えられる.以下では,タスク指向対話で典型的に起こる発話のタイプを時間コストと発話義務の観点から整理し,適当な単位で推論と発話をインタリーブさせることによって上述の枠組がどのように機能するかを考察する.

  241. 人間に対して友好的なインタフェースの実現を目指して : 東京工業大学 田中・徳永研究室 研究室紹介

    田中 穂積, 徳永 健伸, 乾 健太郎

    日本認知科学会大会論文集 = Annual meeting of the Japanese Cognitive Science Society 12 243-244 1995年6月15日

  242. 時間的制約を考慮した実況生成のプランニング

    熊野 正, 徳永 健伸, 乾 健太郎, 田中 穂積

    全国大会講演論文集 50 89-90 1995年3月15日

    詳細を見る 詳細を閉じる

    実況は時間の経過に沿って刻々と変化する状況を自然言語で説明する作業である.実況発話のプランニングでは,実況開始前にプランニングに必要なデータを集めることができないため,事前に完全なプランを組み立てることができない.従って,状況の変化に応じてリアクティブにプランニングを行う必要がある.また聞き手が状況の変化を容易に理解できるように,個々の事象の重要度,状況の局面,各発話の結束性などを考慮する必要がある.本稿では,競馬実況において数値データから発話プランをリアクティブに生成する手法について述べる.

  243. 文章生成システムのためのシステミック文法の開発環境の構築に関する研究

    熊野 正, 徳永 健伸, 乾 健太郎, 田中 穂積

    人工知能学会全国大会論文集 = Proceedings of the Annual Conference of JSAI 8 555-558 1994年6月20日

    ISSN:0914-4293

  244. 確率的制約に基づく発話プランニング

    乾健太郎, 徳永 健伸, 田中 穂積

    自然言語処理 1994 (77) 25-32 1994年

    出版者・発行元:一般社団法人情報処理学会

    詳細を見る 詳細を閉じる

    協調的で効率的な発話のプランニングを実現する上でおこる2つの問題,すなわち,相手のプランがあいまいなときにどう対応するか,相手の推論の予測をどのようにプランニングに埋め込むかに焦点をあて,状況に依存しないヒューリスティクスをつかって発話を選択する枠組について論じる.ここでつかうヒューリスティクスとは,最大の効用をもつ発話を選択するというものである.我々の枠組では,すべての知識を確率的制約として記述し,その制約の上で発話の効用を計算する.本稿では確率的制約の表現と発話の効用の計算についてのべる.

  245. 文章生成における推敲機能の実現について

    乾 健太郎, 徳永 健伸, 田中 穂積

    情報処理学会研究報告自然言語処理(NL) 1992 (93) 17-24 1992年11月19日

    詳細を見る 詳細を閉じる

    文章生成には語彙選択や語順などさまざまな要素に関する選択が必要である.従来のシステムでは,規則を用いて各選択点における選択肢の優先度を局所的に計算するのが一般的だった.しかしながら,選択肢間の優先度の根拠となる評価基準のなかには,評価に将来の選択の予測や大域的な情報を必要とするものもある.それらのすべてを局所的な規則で扱うアーキテクチャは,規則が複雑になりすぎるため,実現や保守の点で望ましくない.本稿では,この問題に対して推敲機能の導入が有効であることについて議論し,推敲機能を備えた生成モデルとその実装方法を提案する.また,その手法を実装した日本語文章生成システムについて述べ,その性能を評価する.To generate good text, many kinds of decisions should be made. Many researchers have been engaged in searching for the heuristics that would make an appropriate decision locally at each decision point. However, even if such heuristics were found, there are still certain kinds of problems that are difficult to consider during the generation process. Those problems can be more easily solved by introducing a revision process after generation. In this paper, we argue the importance of text revision with respect to natural language generation, and propose a computational model of text revisjon. We also discuss its implementation issues and introduce dependency directed back-tracking in order to realize efficient revisions. Finally, we evaluate our method on an experimental Japanese text generation system.

  246. ANLP'92およびIWNLG'92の報告

    徳永 健伸, 乾 健太郎

    情報処理学会研究報告自然言語処理(NL) 1992 (33) 91-98 1992年5月14日

    詳細を見る 詳細を閉じる

    4月1日から3日までイタリアの Trento でおこなわれた第3回応用自然言語処理国際会議 (d Conference on Applied Natural Language Processi) とその直後の4月5日から7日まで Trento から35km離れた castel Ivano という古城でおこなわれた第6回自然言語生成に関する国際ワークショップ (h International Workshop on Natual Language Generati) に参加する機会を得たのでその様子を報告する.This paper reports the outline of the 3rd Conference on Applied Natural Language Processing that was held at Trento, Italy from April 1 to 3, and the 6tn International Workshop on Natural Language Generation that was held at Castel Ivano near Trento following the Applied NLP.

  247. 1980年代の自然言語生成 : 3

    徳永 健伸, 乾 健太郎, Takenobu Tokunaga, Kentaro Inui, 東京工業大学工学部情報工学科, 東京工業大学工学部情報工学科, Dept. of Computer Science Faculty of Eng. Tokyo Institute of Technology, Dept. of Computer Science Faculty of Eng. Tokyo Institute of Technology

    人工知能学会誌 = Journal of Japanese Society for Artificial Intelligence 6 (5) 651-662 1991年9月1日

    ISSN:0912-8085

  248. 1980年代の自然言語生成 : 2

    徳永 健伸, 乾 健太郎, Takenobu Tokunaga, Kentaro Inui

    人工知能学会誌 = Journal of Japanese Society for Artificial Intelligence 6 (4) 510-519 1991年7月1日

    ISSN:0912-8085

  249. 文章生成における推敲の役割

    乾 健太郎, 徳永 健伸, 田中 穂積

    情報処理学会研究報告自然言語処理(NL) 1991 (37) 47-54 1991年5月16日

    詳細を見る 詳細を閉じる

    文章生成には語彙選択や語順などさまざまな要素に関する選択が必要であるが,これらの選択は相互依存の関係にあるため,その実現が難しい.本稿では,この問題へのアプローチとして,一度表層化した文章を繰り返し改良し,最終的に質の高い文章を生成するモデルを提案する.生成過程全体を推敲過程としてとらえることによって,生成に必要な種々の選択を相互に依存する形で実現できることについて議論し,モデルの概要と一部の実現について述べる.In the process of text generation, many kinds of choices should be decided, and these choices are dependent on each other. The versatile architecture which can handle these interdependencies of choices has been the main issue of the text generation research. We propose an architecture in which the text generation process is considered as a revising process. We show our architecture fulfills the above requirement and also describe its partial implementation.

  250. 1980年代の自然言語生成 : 1

    徳永 健伸, 乾 健太郎, Takenobu Tokunaga, Kentaro Inui

    人工知能学会誌 = Journal of Japanese Society for Artificial Intelligence 6 (3) 380-387 1991年5月1日

    ISSN:0912-8085

  251. 推敲に基づく文章生成

    乾 健太郎, 徳永 健伸, 田中 穂積

    全国大会講演論文集 42 124-125 1991年2月25日

    詳細を見る 詳細を閉じる

    文章を生成するには,語乗選択や語順などさまざまな要素に関する決定が必要である.これらの決定は,文章中で述べる話題を選択・構成するwhat-to-sayレベルとwhat-tqsayの内容を表層化するhow-to-sayレベルに分けて考えることができる.2つのレベルの決定は相互に依存するため,その緊密な関係を実現するアーキテクチャの必要性が指摘されている.たとえば,1文の中にどれだけの話題を含めるかという問題は,話題間の意味的なつながりから制約(what-to-sayの制約)を受けると同時に,それを表層化したときに適切な長さの文になるかという制約(how-to-sayの制約)も受ける.また,how-to-sayレベルのみについて考えても,種々の決定が相互に依存し,それらをどの順序で決定すればよいかが必ずしも明らかではない.たとえば,後置詞句の語順は,後置詞句の長さに依存するため,語彙選択を先におこなわなければ適切に決めることができない.語彙選択には照応表現の選択も含まれるが,照応表現は,先行洞と照応詞の距離などに依存するため,適切な照応表現を決定するためには語順の情報が必要である.このように,生成に必要な種々の決定の間には相互依存関係がある.この問題に対する代表的なアプローチの1つに種類の異なる決定を交互におこなう手法があるAppeltやHovyでは,how-to-say決定部が決定の過程で必要に応じてwhat-to-say決定部を呼び出すことにより両者の相互作用を実現しているまた,Hovyは,how-to-say決定過程に対し,決定の種類ごとに異なるモジュールを用意し,モジュールの適用順序を動的に変えることによって,決定の順序に柔軟性を持たせる手法を提案している.しかしながら,これらの手法では,一度決定した要素については変更しないため,将来の影響を十分に予測した上で個々の決定をおこなう必要がある.Appelt,Hovyの手法では,統語的要因を考慮しながらwhat-to-sayを決定するため,what-to-say決定部は複雑なメカニズムを必要とする.また,what-to-say決定部を呼び出すタイミングの管理も困難である文章生成では,論旨展開や照応表現などの文脈的な問題も考慮しなければならないため,メカニズムはさらに複雑になる.本稿では,この問題へのアプローチとして,一度表層化した文章を繰り返し改良し,最終的に質の高い文章を生成するモデルを提案する.一般に,文章を繰り返し改良することを推敲と呼ぶが,生成過程全体を推敲過程としてとらえることによって,生成に必要な種々の決定を相互に依存する形で実現できる.本稿では,推敲に基づく生成モデルの概要と一部の実現について述べる.

  252. 日本語語順の推定モデルとその応用

    徳永 健伸, 岩山 真, 乾 健太郎, 田中 穂積

    情報処理学会研究報告自然言語処理(NL) 1991 (7) 9-16 1991年1月17日

    詳細を見る 詳細を閉じる

    日本語のかかり要素の語順を規定すると考えられる要因のひとつである「かかりの広さ」という概念を用い,日本語単文の次入力を推定するモデルを提案する.「かかりの広さ」は直観的には,かかり要素がかかる動詞に対する限定力と考えることができ,日本語では,かかりの広い要素が先行するとされている.モデルの基本的な考え方は,動詞の結合価バタンから,特定の意味素性と格の組合せをとりうる動詞の数を計算し,この動詞の数でかかりの広さを定量化しようというものである.また,このモデルが音声認識,名詞句の意味の推定,文生成のための語順の基礎的情報として利用できることも述べる.This paper proposes a estimation model of the word order for Japanese simple sentences, which consist of several postpositional phrases followed by a verb. In Japanese word order, the postpositional phrase which has semantically loose relation with the verb, tends to precede in the sentence. As each postpositional phrase is obtained, our model estimates the semantic type and the postposition of the following postpositional phrase by using this tendency. The model also gives a estimation about the verb, which locates in the end of the sentences in Japanese. Our model can be applied to the word sense disambiguation of noun phrases, the speech recognition and the sentence generation.

︎全件表示 ︎最初の5件までを表示

書籍等出版物 7

  1. In Nancy Ide and James Pustejovsky (Eds), Handbook of Linguistic Annotation

    Ryu Iida, Mamoru Komachi, Naoya Inoue, Kentaro Inui, Yuji Matsumoto

    Springer 2017年7月

  2. 批判的思考と市民リテラシー

    乾健太郎

    2016年6月

  3. 自然言語処理における意味・談話情報のコーパスアノテーション. 小川芳樹, 長野明子, 菊地朗 (編), コーパスからわかる言語変化・変異と言語理論, Part V

    福原裕一, 松林優一郎, 乾健太郎

    開拓社 2016年

  4. ワードマップ 批判的思考

    乾健太郎

    2015年1月

  5. Plan, Activity, and Intent Recognition

    Naoya Inoue, Ekaterina Ovchinnikova, Kentaro Inui, nd Jerry R. Hobbs

    2014年3月

  6. 言語と情報科学

    乾健太郎, 関根聡

    2011年7月

  7. 言語情報編集のための広義モダリティ解析に向けて

    乾健太郎, 松吉俊

    2009年9月

︎全件表示 ︎最初の5件までを表示

講演・口頭発表等 268

  1. AIはいつか言葉を理解するか? 〜言語から見る人工知能の現在と未来〜. 招待有り

    乾 健太郎

    秋桜会第339回例会 2019年2月20日

  2. Natural language understanding and assessment. 国際会議

    乾 健太郎

    The 1st Japan-Israel Machine Learning Workshop 2018年11月

  3. 言語から見た人工知能の現在と未来

    乾 健太郎

    朝日地球会議 パネル討論「AI×IoT時代、『人と人の間』はどう変わる」 2018年9月26日

  4. AIはいつか言葉を理解するか? 〜言語から見る人工知能の現在と未来〜

    乾 健太郎

    東北大学111周年萩友会関東交流会 2018年7月22日

  5. Commonsense Reasoning for Natural Language Discourse Understanding.

    乾 健太郎

    Lecture in Machano-informatics, Tokyo University, 2018年7月

  6. 講演

    乾 健太郎

    三菱ケミカルホールディングス Data Scientist Network 2018年度第1Q会合 2018年5月24日

  7. 言葉がわかる人工知能をつくるには 〜自然言語処理の挑戦〜

    乾 健太郎

    宮城第一高校「出前講義」 2018年5月21日

  8. テクノロジー支援と市民参加の可能性

    乾 健太郎, 楊井文人

    ファクトチェック・イニシアティブ設立シンポジウム 2018年4月23日

  9. AI版「赤ペン先生」をつくる!自然言語処理のフロンティア 招待有り

    乾 健太郎

    グロービス経営大学院「知の創造勉強会」 2018年3月26日

  10. 自然言語処理のフロンティア ~自然言語の意味計算とその向こう側~

    乾 健太郎

    Hitachiアカデミックシステム研究会 第41回研究会 2018年3月8日

  11. 言葉がわかる人工知能をつくるには~言葉の不思議と自然言語処理の最前線~

    乾 健太郎

    特別講演会「東北大学in東京 工学部編」 2018年3月

  12. Modeling Commonsense Reasoning for Natural Language Understanding. 国際会議

    乾健太郎

    International Conference for Top and Emerging Computer Scientists (IC-TECS) 2017年12月24日

  13. Modeling Commonsense Reasoning for Natural Language Understanding 国際会議

    乾健太郎

    The 1st NTU-Tohoku U Symposium on Interdisciplinary AI and Human Studies 2017年12月23日

  14. 雛形化による手続き的知識の汎化

    吉成未菜里, 横井祥, 乾健太郎

    情報処理学会第166回データベースシステム研究会ならびに電子情報通信学会データ工学研究会(食メディア研究会協賛) 2017年12月22日

  15. Examining Macro-level Argumentative Structure Features for Argumentative Relation Identification.

    Tatsuki Kuribayashi, Paul Reisert, Naoya Inoue, Kentaro Inui

    第4回自然言語処理シンポジウム・第234回自然言語処理研究会 2017年12月20日

  16. 文法誤り訂正の文単位評価におけるリファレンスレス手法の評価性能

    浅野広樹, 水本智也, 松林優一郎, 乾 健太郎

    第4回自然言語処理シンポジウム・第234回自然言語処理研究会 2017年12月20日

  17. 人工知能の現在と未来

    乾健太郎

    市民のためのサイエンス講座 2017年12月17日

  18. パネル討論

    乾健太郎

    Inter BEE Forum「災害情報とAI~データ・ジャーナリズムの視点から」 2017年11月17日

  19. 記述答案の自動評価に向けて

    乾健太郎

    DiTTシンポジウム「AI時代の教育を考える」 2017年11月9日

  20. カーネル法に基づく共起尺度

    横井祥, 福水健次, 小林颯介, 乾健太郎

    第20回情報論的学習理論ワークショップ (IBIS) 2017年11月8日

  21. . ファクトチェック・イニシアティブ+SmartNews+東北大によるファクトチェック支援技術の実証実験構想

    乾健太郎

    SmartNews メディアパートナーミーティング Vol.5 2017年11月

  22. 発話の語の文体ベクトルの半教師あり学習.

    赤間怜奈, 横井祥, 渡邉研斗, 乾健太郎

    人工知能学会 音声・言語理解と対話処理研究会(SLUD)第81回研究会 第8回対話システムシンポジウム, Vol.2017-SIG-SLUD-B508-27, 2017年10月13日

  23. 深い言語解析のための知識と推論

    乾健太郎

    第3回Language & Robotics 研究会 (LangRobo) 2017年10月7日

  24. 自然言語処理のフロンティア.

    乾健太郎

    Astro-AI 第1回研究会 2017年9月26日

  25. クラウドソーシングを系に組み込んだテキストからの関係知識抽出

    松田耕史, 岡崎直観, 乾健太郎

    第12回NLP若手の会 シンポジウム 2017年9月3日

  26. 自治体QAサービスのためのFAQの自治体間の横断的解析

    伊藤拓海, 鈴木正敏, 田然, 山口健史, 岡崎直観, 乾健太郎

    第12回NLP若手の会 シンポジウム 2017年9月3日

  27. 作業記憶とスクランブリングの関係性

    阿部香央莉, 折田奈甫, 門馬翔太

    第12回NLP若手の会 シンポジウム 2017年9月3日

  28. トピックに関する因果関係知識を利用した賛否分類

    塙一晃, 佐々木彬, 岡崎直観, 乾健太郎

    第12回NLP若手の会 シンポジウム 2017年9月3日

  29. マイクロブログを利用した評判分析に向けてのデータセット作成

    栗原理聡, 水本智也, 乾健太郎

    第12回NLP若手の会 シンポジウム 2017年9月3日

  30. 日本語メタファー表現に対する分散表現の評価と考察

    菊地凜, 松林優一郎, 乾健太郎

    第12回NLP若手の会 シンポジウム 2017年9月3日

  31. 発話スタイル空間の教師なし学習およびスタイル制御可能な対話システムの実現

    赤間怜奈, 渡邉研斗, 横井祥, 乾健太郎

    第12回NLP若手の会 シンポジウム 2017年9月3日

  32. カーネル密度推定に基づく関係予測

    横井祥, 乾健太郎

    第12回NLP若手の会 シンポジウム 2017年9月3日

  33. 数量表現と比較に着目した意味解析に向けて

    佐々木翔大, 田然, 乾健太郎

    第12回NLP若手の会 シンポジウム 2017年9月3日

  34. Dependency-based Compositional Semantics Vectorに基づく文脈付き選択選好モデル

    高橋諒, 松林優一郎, 中山周, 田然, 乾健太郎

    第12回NLP若手の会 シンポジウム 2017年9月3日

  35. Wikipedia を知識源に用いた文書検索と読解によるクイズ解答システム

    鈴木正敏, 松田耕史, 岡崎直観, 乾健太郎

    第12回NLP若手の会 シンポジウム 2017年9月3日

  36. 分散表現から感情極性は予測できるか?

    中村拓, 田然, 松田耕史, 乾健太郎

    第12回NLP若手の会 シンポジウム 2017年9月3日

  37. 大学入試の国語の記述式問題の自動採点に向けて

    水本智也, 乾健太郎

    第12回NLP若手の会 シンポジウム 2017年9月3日

  38. 文法誤り訂正のリファレンスレス評価における文単位評価

    浅野広樹, 水本智也, 乾健太郎

    第12回NLP若手の会 シンポジウム 2017年9月3日

  39. コノテーションに基づいた名詞の感情極性の予測

    白井穂乃, 田然, 松田耕史, 乾健太郎

    第12回NLP若手の会 シンポジウム 2017年9月3日

  40. 人工知能・自然言語処理から見える産業のスマート化 〜人工知能ブームへの備え方〜.

    乾健太郎

    東北生産性本部平成29年度労使セミナー 2017年9月

  41. 言葉がわかる人工知能をつくるには 〜自然言語処理の挑戦〜

    乾健太郎

    河合塾仙台校「知の広場」 2017年7月27日

  42. 言葉がわかる人工知能をつくるには 〜言葉の不思議と自然言語処理の最前線〜

    乾健太郎

    東北大学オープンキャンパス模擬講義 2017年7月25日

  43. Modelling Relations between Objects for Referring Expression Comprehension

    冉 文升, 田然, 岡崎直観, 乾健太郎

    第232回自然言語処理研究会 2017年7月19日

  44. Commonsense Reasoning for Natural Language Discourse Understanding 国際会議

    Kentaro Inui

    Machano-informatics 2017年7月

  45. Commonsense Reasoning for Natural Language Discourse Understanding

    乾健太郎

    Lecture in Machano-informatics, Tokyo University, 2017年7月

  46. 言葉がわかる人工知能をつくるには 〜自然言語処理の挑戦〜

    乾健太郎

    渋谷教育学園幕張高等学校 2017年7月

  47. 言葉がわかる人工知能をつくるには 〜自然言語処理の挑戦〜

    乾健太郎

    宮城県立富谷高校 2017年6月

  48. 独立性尺度に基づく知識の粒度の教師なし推定

    横井祥, 持橋大地, 高橋諒, 岡崎直観, 乾健太郎

    第31回人工知能学会全国大会 2017年5月23日

  49. メロディと歌詞の相関に基づく自動歌詞生成

    渡邉研斗, 松林優一郎, 深山覚, 中野倫靖, 後藤真孝, 乾健太郎

    第231回自然言語処理研究会・第116回音声言語情報処理研究会 2017年5月15日

  50. 複単語表現を利用した因果関係推定モデルの改善

    佐々木翔大, 高瀬翔, 井之上直也, 岡崎直観, 乾健太郎

    第231回自然言語処理研究会・第116回音声言語情報処理研究会 2017年5月15日

  51. 言葉がわかる人工知能をつくるには 〜自然言語処理の挑戦〜

    乾健太郎

    宮城第一高校 2017年5月

  52. 従属性に基づく事態間関係知識の粒度調整

    横井祥, 持橋大地, 岡崎直観, 乾健太郎

    言語処理学会第23回年次大会 2017年3月13日

  53. 人工知能・自然言語処理から見える産業のスマート化:人工知能ブームへの備え方

    乾健太郎

    仙台商工会議所 工業部会2 月常任委員会講演会 2017年2月10日

  54. 自然言語処理のフロンティア「新たな数理科学の可能性に向けて:IoT・人工知能・量子ウォークとその周辺」

    乾健太郎

    横浜国立大学研究集会 2016年12月18日

  55. カーネル法を用いた従属性の高い部分木ペアの教師なし推定

    横井祥, 持橋大地, 岡崎直観, 乾健太郎

    第19回情報論的学習理論ワークショップ 2016年11月17日

  56. 文章の「行間」を読むコンピュータの実現を目指して

    乾健太郎

    日立ソリューションズ東日本事業化発表会 2016年11月10日

  57. Commonsense Reasoning for Natural Language Discourse Understanding 国際会議

    乾健太郎

    Bilateral Workshop between Tohoku University and National Tsing Hua University 2016年10月25日

  58. 「行間を読む」言語理解のための知識と推論

    乾健太郎, 関根聡

    第4回けいはんなエジソンの会 2016年10月3日

  59. Computational models of commonsense reasoning for natural language discourse understanding 国際会議

    乾健太郎

    AEARU Web Technology and Computer Science Workshop 2016 2016年9月20日

  60. 自然言語処理における深層ニューラルネットワーク

    岡崎直観

    Prometech Simulation Conference 2016 2016年9月9日

  61. 言語処理における常識的知識の獲得・活用

    岡崎直観

    第32回ファジィシステムシンポジウム 2016年8月31日

  62. seq2seqモデルに基づく略語の自動生成

    鶴田崇仁, 松田耕史, 岡崎直観, 乾健太郎

    第11回NLP若手の会 シンポジウム 2016年8月

  63. 質問応答タスクの設定と文章読解モデルの比較・検討

    清野舜, 岡崎直観, 乾健太郎

    第11回NLP若手の会 シンポジウム 2016年8月

  64. 独立性判定基準に基づく知識の汎化

    横井祥, 持橋大地, 岡崎直観, 乾健太郎

    第11回NLP若手の会 シンポジウム 2016年8月

  65. 分散表現を用いた因果関係のモデル化

    佐々木翔大, 高瀬翔, 井之上直也, 岡崎直観, 乾健太郎

    第11回NLP若手の会 シンポジウム 2016年8月

  66. 地理情報に対する実体験に焦点を当てた意見抽出に向けて

    栗原理聡, 水本智也, 松田耕史, 乾健太郎

    第11回NLP若手の会 シンポジウム 2016年8月

  67. 共参照関係に基づく分散表現の共有と動的更新

    小林颯介, 岡崎直観, 乾健太郎

    第11回NLP若手の会 シンポジウム 2016年8月

  68. 自然言語処理との付き合い方 〜災害対応とメディアを例に〜

    乾健太郎

    日本原子力学会HMS研究部会2016年度夏期セミナー 2016年7月

  69. ツイート中の地理情報に対する時間的極性の自動推定

    珊瑚 彩主紀, 松田 耕史, 岡崎 直観, 乾 健太郎

    人工知能学会第30回全国大会 2016年6月

  70. 因果関係に基づく雑談対話発話生成の試み

    佐藤 祥多, 井之上 直也, 乾 健太郎, 渡部 生聖, 樋口 佐和

    人工知能学会第30回全国大会 2016年6月

  71. ストーリー展開と一貫性を同時に考慮した歌詞生成モデル

    渡邉 研斗, 松林 優一郎, 乾 健太郎, 深山 覚, 中野 倫靖, 後藤 真孝

    人工知能学会第30回全国大会 2016年6月

  72. 交通オントロジーと説明生成に基づく交通危険予測

    高橋 諒, 井之上 直也, 栗谷 康隆, 山本 風人, 乾 健太郎

    人工知能学会第30回全国大会 2016年6月

  73. Businnovare - PwC Technology Day -

    乾健太郎

    パネリスト 2016年6月

  74. 関係代数の意味論に基づく含意関係認識

    田 然, 岡崎直観, 乾健太郎

    人工知能学会第30回全国大会 2016年6月

  75. 言葉を理解する人工知能のいま 〜自然言語処理・人工知能から見える社会・産業のスマート化〜

    乾健太郎

    産学連携セミナー「第102回寺子屋せんだい」 2016年5月

  76. 「拡張固有表表現+Wikipedia」データ

    関根聡, 安藤まや, 松田耕史, 鈴木正敏, 乾健太郎

    言語処理学会第22回年次大会 2016年3月

  77. 利用物に関する知識のコーパスアノテーション

    上村明衣, 折田奈甫, 岡崎直観, 乾健太郎

    言語処理学会第22回年次大会 2016年3月

  78. 画像説明文生成に向けた物体間の関係の認識

    村岡雅康, Sumit Maharjan, 齋藤真樹, 山口光太, 岡崎直観, 岡谷貴之, 乾健太郎

    言語処理学会第22回年次大会 2016年3月

  79. 日本語Wikificationコーパスの構築に向けて

    Davaajav Jargalsaikhan, 岡崎直観, 松田耕史, 乾健太郎

    言語処理学会第22回年次大会 2016年3月

  80. Wikipedia記事に対する拡張固有表現ラベルの多重付与

    鈴木正敏, 松田耕史, 関根聡, 岡崎直観, 乾健太郎

    言語処理学会第22回年次大会 2016年3月

  81. An Ontology for Question-Answering on Minecraft

    Dumont Corentin, 田然, 乾健太郎

    言語処理学会第22回年次大会 2016年3月

  82. 分散表現による文脈情報を用いた選択選好モデル

    大野雅之, 井之上直也, 松林優一郎, 岡崎直観, 乾健太郎

    言語処理学会第22回年次大会 2016年3月

  83. 感情状態に基づく因果関係推論の一般化

    井之上直也, 岡崎直観, 乾健太郎

    言語処理学会第22回年次大会 2016年3月

  84. 談話内における局所文脈の動的分散表現

    小林颯介, 田然, 岡崎直観, 乾健太郎

    言語処理学会第22回年次大会 2016年3月

  85. 自然言語処理を活用した防災情報の整理と共有

    乾健太郎

    仙台防災未来フォーラム2016「『しなやかな社会』の情報通信ネットワークを目指して」 2016年3月

  86. もしも災害で情報が混乱してしまったら 〜自然言語処理を活用した防災情報の整理と共有〜

    乾健太郎

    東北大学大学院情報科学研究科シンポジウム「『情報科学』から『防災』を考える」 2016年3月

  87. 分散表現に基づく選択選好モデルの文脈化

    大野雅之, 井之上直也, 松林優一郎, 岡崎直観, 乾健太郎

    情報処理学会研究報告. 自然言語処理研究会報告 2016年1月

  88. 自然言語処理・人工知能から見るビッグデータの今と未来

    乾健太郎

    第1回 ビッグデータ車座 in 仙台 2016年1月

  89. 人工知能研究から見えるスマート社会の地平

    乾健太郎

    東北大学電気・情報系東京フォーラム 2015年11月

  90. 言葉がわかる人工知能をつくる 〜自然言語処理の挑戦〜

    乾健太郎

    仙台一高 2015年10月

  91. 言葉がわかるコンピュータをつくる 〜言葉の不思議と自然言語処理の最前線〜

    乾健太郎

    東北大学祭模擬講義 2015年10月

  92. 大規模知識獲得で深化する自然言語処理のフロンティア

    乾健太郎

    モバイル・コンテンツ・フォーラム「モバイルビジネスの発展を実現する人工知能の活用とは」 2015年10月

  93. 言葉がわかる人工知能をつくる 〜自然言語処理の挑戦〜

    乾健太郎

    宮城一高 2015年10月

  94. Determining Argument Strength and Consistency using a Causality-based Knowledge Graph

    Paul Reisert, Naoya Inoue, Naoaki Okazaki, Kentaro Inui

    第10回NLP若手の会 シンポジウム 2015年9月

  95. ストーリー展開を考慮した自動歌詞生成システム

    渡邉研斗, 松林優一郎, 乾健太郎, 深山覚, 中野倫靖, 後藤真孝

    第10回NLP若手の会 シンポジウム 2015年9月

  96. Wikipediaエントリの拡張固有表現階層への自動分類

    鈴木正敏, 松田耕史, 関根聡, 岡崎直観, 乾健太郎

    第10回NLP若手の会 シンポジウム 2015年9月

  97. Exploring the Challenges of Entity Linking in Knowledge-based Question Answering

    周双双, 岡崎直観, 乾健太郎

    第10回NLP若手の会 シンポジウム 2015年9月

  98. 審査委員としての視点による説明

    乾健太郎

    平成27年度科研費学内説明会 2015年9月

  99. 言語コーパスへの重層的意味情報付与 〜自然言語処理から見たコーパス分析〜

    乾健太郎, 松林優一

    言語変化・変異研究ユニット第二回ワークショップ「コーパスからわかる言語の可変性と普遍性」 2015年9月

  100. 自然言語処理による情報編集

    乾健太郎

    SmartNews Tech Night Vol.3 2015年8月

  101. 「行間を読む」自然言語処理への挑戦 〜知識、学習、推論、そしてグラウンディング〜

    乾健太郎

    第18回画像の認識・理解シンポジウム (MIRU) 2015年7月

  102. 文間弱対立関係認識のためのNatural Logicの拡張

    大江貴裕, 水野淳太, 稲田和明, 乾健太郎

    人工知能学会第29回全国大会 2015年6月

  103. ディベート人工知能における意見生成

    柳井孝介, 三好利昇, 柳瀬利彦, 佐藤美沙, 丹羽芳樹, Reisert Paul, 乾健太郎

    人工知能学会第29回全国大会 2015年6月

  104. Bi-gram連接表と単語列変形規則に基づく回文自動生成

    清野舜, 渡邉研斗, 岡崎直観, 乾健太郎

    人工知能学会第29回全国大会 2015年6月

  105. 関係知識獲得のための意味表現の学習

    高瀬翔, 岡崎直観, 乾健太郎

    人工知能学会第29回全国大会 2015年5月

  106. Twitterを利用した地域毎の要望抽出

    栗原理聡, 佐々木彬, 松田耕史, 岡崎直観, 乾健太郎

    人工知能学会第29回全国大会 2015年5月

  107. 物理モデルと論理推論の統合による運転シーンの潜在的危険の予測

    小林颯介, 井之上直也, 栗谷康隆, 近藤敏之, 安部克則, 奥野英一, 乾健太郎

    自動車技術会 2015年春季大会学術講演会講演予稿集 2015年5月

  108. 単語分散表現のshift-reduce型構文解析への利用

    小松 広弥, 田 然, 岡崎 直観, 乾 健太郎

    情報処理学会研究報告 自然言語処理(NL) 2015年3月

  109. キーワードの自動拡張に基づくイベント言及ツイートの収集

    五十嵐祐貴, 大野雅之, 岡崎直観, 乾健太郎

    言語処理学会第21回年次大会 2015年3月

  110. Exploring Linguistic Features for Cross-document Named Entity Disambiguation

    周双双, Canasai Kruengkrai, 乾健太郎

    言語処理学会第21回年次大会 2015年3月

  111. 言語処理のための仮説推論エンジン Phillip

    山本風人, 井之上直也, 乾健太郎

    言語処理学会第21回年次大会 2015年3月

  112. 構成性に基づく関係パタンの意味計算

    高瀬翔, 岡崎直観, 乾健太郎

    言語処理学会第21回年次大会 2015年3月

  113. Computationalizing a Toulmin Model for Argumentation Generation

    Paul Reisert, Naoya Inoue, Kentaro Inui, Toshihiko Yanase, Kohsuke Yanai

    言語処理学会第21回年次大会 2015年3月

  114. 日本語述語項構造解析のための統語パターン分析

    中山周, 松林優一郎, 乾健太郎

    言語処理学会第21回年次大会 2015年3月

  115. 機能表現に基づく日本語事実性解析

    成田和弥, 水野淳太, 上岡裕大, 菅野美和, 乾健太郎

    言語処理学会第21回年次大会 2015年3月

  116. 形式意味論に基づく出来事間関係認識に向けて-リソース構築の展望とテンス「タ」のアノテーション

    宇津木舞香, 稲田和明, 金子貴美, 戸次大介, 乾健太郎

    言語処理学会第21回年次大会 2015年3月

  117. Learning Visual Attributes from Image and Text

    Maharjan Sumit, 齋藤真樹, 山口光太, 岡崎直観, 岡谷貴之, 乾健太郎

    言語処理学会第21回年次大会 2015年3月

  118. 大規模歌詞データからの潜在的トピック遷移のモデル化

    渡邉研斗, 松林優一郎, 乾健太郎, 後藤真孝

    情報処理学会 第77回全国大会予稿集 2015年3月

  119. 日本語文における機能表現意味ラベル付与と事実性解析への効果

    上岡裕大, 成田和弥, 菅野美和, 水野淳太, 乾健太郎

    情報処理学会 第77回全国大会予稿集 2015年3月

  120. 日本語文内で表現されるイベント間の時間的な順序関係の認識

    稲田和明, 松林優一郎, 乾健太郎

    情報処理学会 第77回全国大会予稿集 2015年3月

  121. 自然言語処理から見える知能情報技術の近未来

    乾健太郎

    JASA東北 知能情報フェスタ 2015年3月

  122. ビッグデータで加速する自然言語処理

    乾健太郎

    東北IT推進機構 勉強会 2015年3月

  123. 自然言語処理の立場から見た近未来

    乾健太郎

    情報処理学会第77回全国大会「トップランナーが語るITの未来」 2015年3月

  124. Modeling "Reading between the Lines" Based on Scalable and Trainable Abduction and Large-scale Knowledge Acquisition

    Kentaro Inui

    Workshop on Deep and Large-Scale Semantic Processing: The Way Ahead 2015年3月

  125. 計算機によるデータに基づく言い換え

    乾健太郎

    第6回産業日本語研究会・シンポジウム 2015年2月

  126. 場所参照表現タグ付きコーパスの構築と評価

    松田耕史, 佐々木 彬, 岡崎直観, 乾健太郎

    情報処理学会研究報告 自然言語処理(NL) 2015年1月

  127. Modeling "Reading between the Lines" Based on Scalable and Trainable Abduction and Large-scale Knowledge Acquisition

    Kentaro Inui

    NII Shonan Meeting Seminar 057: Towards Explanation Production Combining Natural Language Processing and Logical Reasoning 2014年11月

  128. 言葉がわかるコンピュータを創るには 〜言葉の不思議と自然言語処理の最前線〜

    乾健太郎

    東北大学祭模擬講義 2014年11月

  129. 実用的な自動歌詞生成に向けて

    渡邉研斗, 松林優一郎, 乾健太郎, 後藤真孝

    第9回NLP若手の会 シンポジウム 2014年9月

  130. 述部機能表現に対する意味ラベル付与と事実性解析への適用

    上岡裕大, 菅野美和, 成田和弥, 水野淳太, 乾健太郎

    第9回NLP若手の会 シンポジウム 2014年9月

  131. 共参照解析のための事象間関係知識の一般化に向けて

    井之上直也, 杉浦純, Canasai Kruengkrai, 乾健太郎

    第9回NLP若手の会 シンポジウム 2014年9月

  132. Word2vecの並列実行時の学習速度の改善

    岡崎直観, 乾健太郎

    情報処理学会研究報告 自然言語処理(NL) 2014年7月

  133. 近似頻度計測手法を用いた大規模データからの関係獲得

    高瀬翔, 岡崎直観, 乾健太郎

    情報処理学会研究報告 自然言語処理(NL) 2014年7月

  134. 述部機能表現に対する意味ラベル付与

    上岡裕大, 成田和弥, 水野淳太, 乾健太郎

    情報処理学会 第216回自然言語処理研究会・第101回 音声言語情報処理研究会 2014年5月

  135. インターネット上の当選運動・落選運動の分析

    船木洋晃, 佐々木彬, 岡崎直観, 乾健太郎, 深田陽介, 竹下隆一郎, 田森秀明, 野澤博

    人工知能学会第28回全国大会 2014年5月

  136. ビッグデータ時代の自然言語処理

    乾健太郎

    テレコムサービス協会東北支部セミナー 2014年5月

  137. 知のメディアへ:自然言語処理からの展開

    乾健太郎

    人工知能学会全国大会特別セッション「人間と調和した創造的協働を実現する知的情報処理システム」 2014年5月

  138. 曖昧になる境界、散り散りになる課題

    乾健太郎

    言語処理学会第20回年次大会ワークショップ「自然言語処理の発展に向けた情報共有・討論」 2014年3月

  139. 高速な類似度計算手法による関係パタンのクラスタリング

    高瀬翔, 岡崎直観, 乾健太郎

    言語処理学会第20回年次大会 2014年3月

  140. 場所参照表現のグラウンディングに向けて

    佐々木彬, 五十嵐祐貴, 渡邉陽太郎, 乾健太郎

    言語処理学会第20回年次大会 2014年3月

  141. 同一文内の表現対を対象とした日本語における時間関係認識

    稲田和明, 松林優一郎, 乾健太郎

    言語処理学会第20回年次大会 2014年3月

  142. 統計的日本語述語項構造解析のための素性設計再考

    松林優一郎, 乾健太郎

    言語処理学会第20回年次大会 2014年3月

  143. An Investigation of Evidence Relations within Social Media Conversations

    Paul Reisert, Junta Mizuno, Miwa Kanno, Naoaki Okazaki, Kentaro Inui

    言語処理学会第20回年次大会 2014年3月

  144. 大局的な構造を考慮した歌詞自動生成システムの提案

    渡邉研斗, 松林優一郎, 乾健太郎, 後藤真孝

    言語処理学会第20回年次大会 2014年3月

  145. 共参照解析における事象間関係知識の適用

    杉浦純, 井之上直也, 乾健太郎

    言語処理学会第20回年次大会 2014年3月

  146. 共参照解析のための事象間関係知識の文脈化

    井之上直也, 杉浦純, 乾健太郎

    言語処理学会第20回年次大会 2014年3月

  147. 実世界指向情報構造化支援のための情報抽出技術

    渡邉陽太郎, 佐々木彬, 五十嵐祐貴, 岡崎直観, 乾健太郎

    言語処理学会第20回年次大会 2014年3月

  148. ガウス分布による単語と句の意味の分布的表現

    島岡聖世, 村岡雅康, 山本風人, 渡邉陽太郎, 岡崎直観, 乾健太郎

    言語処理学会第20回年次大会 2014年3月

  149. 係り受け関係を用いた句ベクトルの生成

    村岡雅康, 島岡聖世, 山本風人, 渡邉陽太郎, 岡崎直観, 乾健太郎

    言語処理学会第20回年次大会 2014年3月

  150. ソーシャルメディア時代のソーシャルリスニング

    乾健太郎

    仙台泉倫理法人会モーニングセミナー 2014年1月

  151. 情報信頼性の検証への自然言語処理的アプローチ

    乾健太郎, 岡崎直観

    人工知能学会第69回言語・音声理解と対話処理研究会(SIG-SLUD) 2013年12月

  152. ビッグデータから知をつむぐ自然言語処理

    乾健太郎

    第13回東北大学-NTT技術交流会 2013年12月

  153. 日本語述語項構造アノテーションに関わる諸問題の分析

    松林優一郎, 飯田龍, 笹野遼平, 横野光, 松吉俊, 藤田篤, 宮尾祐介, 乾健太郎

    情報処理学会 第214回自然言語処理研究会 2013年11月

  154. ツイッター分析に基づく福島県産桃に対する風評の実態解明とその対策

    岡崎直観, 佐々木彬, 乾健太郎, 阿部博史, 石田望

    第26回日本リスク研究学会年次大会 2013年11月

  155. 言葉が少しわかるようになったコンピュータの使い方

    乾健太郎

    仙台エコーライオンズクラブ 2013年11月

  156. ビッグデータから知をつむぐ自然言語処理

    乾健太郎

    東北大学 電気・情報 東京フォーラム 2013年11月

  157. ビッグデータと対話する自然言語処理

    乾健太郎

    お茶の水女子大学情報科学科 理学総論 2013年11月

  158. 自然言語処理から見たビッグデータの可能性

    乾健太郎

    ITCみやぎ・SAAJ東北・JISTA東北ワークショップ2013 2013年10月

  159. 「行間を読む」自然言語処理への挑戦

    乾健太郎

    豊田工業大学第9回ジョイントCSセミナー 2013年10月

  160. ソーシャルメディア時代のソーシャルリスニング

    乾健太郎

    仙台青葉倫理法人会モーニングセミナー 2013年10月

  161. 防災情報データベース化支援システム

    渡邉陽太郎, 乾健太郎

    ALAGIN & NLP若手の会 合同シンポジウム 2013年9月

  162. Acquisition of Inference Rules for Distinct Relative Entities

    Daiqin, Naoya Inoue, Kentaro Inui, Naoaki Okazaki

    ALAGIN & NLP若手の会 合同シンポジウム 2013年9月

  163. 事実性解析における否定のスコープ解析

    成田和弥, 水野淳太, 乾健太郎

    ALAGIN & NLP若手の会 合同シンポジウム 2013年9月

  164. ツイッターからの社会の「論点」を探る

    渡邉研斗, 岡崎直観, 乾健太郎

    ALAGIN & NLP若手の会 合同シンポジウム 2013年9月

  165. 拡張モダリティ解析器の試作と課題分析

    水野淳太, 成田和弥, 乾健太郎, 大竹清敬, 鳥澤健太郎

    ALAGIN & NLP若手の会 合同シンポジウム 2013年9月

  166. 耐災害ICTを目指す自然言語処理

    乾健太郎

    東北大学電気通信研究機構シンポジウム 2013年7月

  167. 自然言語処理によるネット情報分析、およびツイート分析と報道

    乾健太郎, 岡崎直観

    三金会 2013年7月

  168. ビッグデータと対話する自然言語処理

    乾健太郎

    東京大学大学院情報理工学研究科コンピュータ科学専攻講演会 2013年7月

  169. 機械学習に基づくマイクロブログ上のテキストの正規化

    佐々木彬, 水野淳太, 岡崎直観, 乾健太郎

    人工知能学会第27回全国大会 2013年6月

  170. マイクロブログにおける同意・反論関係を用いた情報信頼度推定

    佐藤雅宏, 岡崎直観, 乾健太郎

    人工知能学会第27回全国大会 2013年6月

  171. ウェブ文書の構造を利用した場所名・住所ペアの獲得

    佐藤貴大, 岡崎直観, 乾健太郎

    人工知能学会第27回全国大会 2013年6月

  172. 文章構造を用いたWhy型質問応答システム

    車智修, 鍋島啓太, 水野淳太, 岡崎直観, 乾健太郎

    人工知能学会第27回全国大会 2013年6月

  173. ビッグデータ時代の自然言語処理

    乾健太郎

    東北データベースソサエティ 2013年6月

  174. 大学でワクワクするためのいくつかのヒント 〜言葉がわかるコンピュータの話とともに〜

    乾健太郎

    仙台第三高等学校 2013年5月

  175. 語彙知識と構成性に基づく日本語事実性解析

    成田和弥, 水野淳太, 乾健太郎

    言語処理学会第19回年次大会 2013年3月

  176. Twitter上での誤情報と訂正情報の自動分類

    渡邉研斗, 鍋島啓太, 岡崎直観, 乾健太郎

    言語処理学会第19回年次大会 2013年3月

  177. マイクロブログからの誤情報の発見と集約

    鍋島啓太, 水野淳太, 岡崎直観, 乾健太郎

    言語処理学会第19回年次大会 2013年3月

  178. ソーシャルメディア上の発言とユーザー間の関係を利用した批判的ユーザーの抽出

    高瀬翔, 村上明子, 榎美紀, 岡崎直観, 乾健太郎

    言語処理学会第19回年次大会 2013年3月

  179. 数量の大小の自動判定:「彼は身長が2mある」は高いか低いか

    成澤克麻, 渡邉陽太郎, 水野淳太, 岡崎直観, 乾健太郎

    言語処理学会第19回年次大会 2013年3月

  180. 効率的な推論処理のための日本語文の論理式変換に向けて

    稲田和明, 松林優一郎, 井之上直也, 乾健太郎

    言語処理学会第19回年次大会 2013年3月

  181. オートエンコーダにおける単語ベクトルの学習

    島岡聖世, 山本風人, 乾健太郎

    言語処理学会第19回年次大会 2013年3月

  182. 文節間限定関係に基づく文間弱対立関係認識

    大西真輝, 水野淳太, 福原裕一, 渡邉陽太郎, 乾健太郎

    言語処理学会第19回年次大会 2013年3月

  183. Exploiting Dependency Context Gazetteers for Named Entity Recognition

    Han-Cheol Cho, Naoaki Okazaki, Kentaro Inui

    言語処理学会第19回年次大会 2013年3月

  184. ネット情報の信頼性分析を支援する言語処理技術

    乾健太郎

    情報処理学会 第75回全国大会 2013年3月

  185. Twitterにおける誤情報の拡散収束過程の可視化

    渡邉研斗, 鍋島啓太, 水野淳太, 岡崎直観, 乾健太郎

    情報処理学会 第75回全国大会予稿集 2013年3月

  186. 訂正パターンに基づく誤情報の抽出と集約

    鍋島啓太, 水野淳太, 岡崎直観, 乾健太郎

    情報処理学会 第75回全国大会予稿集 2013年3月

  187. マイクロブログユーザからの現地被災者抽出の技術的支援

    水野淳太, 岡崎直観, 乾健太郎

    情報処理学会 第75回全国大会予稿集 2013年3月

  188. 返信・非公式リツイートで表明される論述関係の認識

    大和田裕亮, 岡崎直観, 乾健太郎, 石塚満

    情報処理学会 第75回全国大会予稿集 2013年3月

  189. 言葉がわかるコンピューターはどこまでできたか 〜言葉の不思議と自然言語処理の最前線

    乾健太郎

    東北大学サイエンスカフェ 2013年2月

  190. ビッグデータ時代の自然言語処理

    乾健太郎

    鹿児島大学情報工学科先端科学特別講義 2013年2月

  191. 「被災者」ツィートに見る東日本大震災発生1週間の災害対応過程の分析

    佐藤翔輔, 今村文彦, 水野淳太, 岡崎直観, 乾健太郎

    平成24年度東北地域自然災害科学研究集会 2012年12月

  192. Online Large-margin Weight Learning for First-order Logic-based Abduction

    Naoya Inoue, Kazeto Yamamoto, Yotaro Watanabe, Naoaki Okazaki, Kentaro Inui

    In Proceedings of the 15th Information-Based Induction Sciences Workshop 2012年11月

  193. 言葉が少しわかるようになったコンピュータの使い方

    乾健太郎

    仙台はなもく七三会 2012年11月

  194. 重み付き仮説推論における部分的な正解仮説からの識別学習

    山本風人, 井之上直也, 渡邉陽太郎, 岡崎直観, 乾健太郎

    NLP若手の会 第7回シンポジウム 2012年9月

  195. 命題ネット構築にむけて

    鍋島啓太, 水野淳太, 渡邉陽太郎, 岡崎直観, 乾健太郎

    NLP若手の会 第7回シンポジウム 2012年9月

  196. 単語間関係に基づく含意関係認識

    水野淳太, 隅田飛鳥, 乾健太郎

    NLP若手の会 第7回シンポジウム 2012年9月

  197. 談話関係認識への連想情報の応用

    杉浦純, 井之上直也, 乾健太郎

    NLP若手の会 第7回シンポジウム 2012年9月

  198. 名詞カテゴリからの関係知識獲得に向けて

    高瀬翔, 岡崎直観, 乾健太郎

    NLP若手の会 第7回シンポジウム 2012年9月

  199. Extending ILP-based Abductive Inference with Cutting Plane Inference

    Naoya Inoue, Kentaro Inui

    In IPSJ SIG Technical Reports 2012年9月

  200. Natural Logicと条件付確率場の融合による構成的文間関係認識

    渡邉陽太郎, 水野淳太, 岡崎直観, 乾健太郎

    情報処理学会研究報告 2012年9月

  201. ネット情報の信頼性分析を支援する言語処理技術

    乾健太郎

    第86回デジタルドキュメント研究会 2012年7月

  202. 誤差逆伝播を利用した重み付き仮説推論の教師あり学習

    山本風人, 井之上直也, 渡邊陽太郎, 岡崎直観, 乾健太郎

    情報処理学会研究報告 2012年5月

  203. 数量表現を伴う文における含意関係認識の課題分析

    成澤克麻, 渡邉陽太郎, 水野淳太, 岡崎直観, 乾健太郎

    言語処理学会第18回年次大会論文集 2012年3月

  204. Web文書からの人の安全・危険に関わる情報の抽出

    岡崎直観, 成澤克麻, 乾健太郎

    言語処理学会第18回年次大会論文集 2012年3月

  205. 英作文支援のための用例検索システムの構築

    高松優, 水野淳太, 岡崎直観, 乾健太郎

    言語処理学会第18回年次大会論文集 2012年3月

  206. 冠詞誤り訂正時における訂正根拠の提示

    梅澤次郎, 水野淳太, 岡崎直観, 乾健太郎

    言語処理学会第18回年次大会論文集 2012年3月

  207. 隠れ変数を持つ識別モデルによる文間意味関係の学習

    渡邉陽太郎, 水野淳太, 岡崎直観, 乾健太郎

    言語処理学会第18回年次大会論文集 2012年3月

  208. 意味カテゴリの階層関係を活用した集合拡張

    高瀬翔, 岡崎直観, 乾健太郎

    言語処理学会第18回年次大会論文集 2012年3月

  209. 説明生成に基づく談話構造解析の課題分析

    杉浦純, 井之上直也, 乾健太郎

    言語処理学会第18回年次大会論文集 2012年3月

  210. 大規模世界知識を用いた仮説推論による談話解析の課題と対策

    井之上直也, 乾健太郎, Ekaterina Ovchinnikova, Jerry R Hobbs

    言語処理学会第18回年次大会論文集 2012年3月

  211. 大規模言語資源時代の意味談話処理

    乾健太郎

    言語処理学会第18回年次大会チュートリアル講演 2012年3月

  212. Toward deep processing of language in the era of large-scale knowledge resources: Time for formal semantics to meet NLP again

    Kentaro Inui

    In Proceedings of the 8th International Workshon on Logic and Engineering of Natural Language Semantics (LENLS8) 2011年12月

  213. 日本語事実性解析課題の経験的分析

    成田和弥, 水野淳太, 乾健太郎

    情報処理学会研究報告 第204回自然言語処理研究会 2011年11月

  214. 事実性解析のその後

    成田和弥, 乾健太郎

    機能表現・文末表現に関するワークショップ 2011年10月

  215. Scalable Abduction for Deep NLP

    Kentaro Inui, Naoya Inoue

    Workshop 'Perspectives on Processing Japanese' 2011年10月

  216. 日本語事実性解析課題の経験的分析

    成田和弥, 水野淳太, 乾健太郎

    NLP若手の会 第6回シンポジウム 2011年9月

  217. 大規模語彙知識を用いた仮説推論による文章理解モデルの構築に向けて

    杉浦純, 井之上直也, 乾健太郎

    NLP若手の会 第6回シンポジウム 2011年9月

  218. Evidence Evaluation Using Argument Schemes

    Eric Nichols, Kentaro Inui

    NLP若手の会 第6回シンポジウム 2011年9月

  219. 文間関係認識における意味的関係の分析

    水野淳太, 乾健太郎

    NLP若手の会 第6回シンポジウム 2011年9月

  220. An ILP Formulation of Abductive Inference for Discourse Interpretation

    Naoya Inoue, Kentaro Inui

    In IPSJ SIG Technical Reports 2011年9月

  221. Scalable Abduction for Deep NLP

    Kentaro Inui, Naoya Inoue

    The Shonan Meeting on Knowledge-leveraged Computational Thinking through Natural Language Processing and Statistical Logic 2011年9月

  222. 大規模言語資源に基づくWeb情報の組織化

    乾健太郎

    ALAGINシンポジウム2011 2011年9月

  223. ネット情報の信頼性を検証するための技術的支援

    乾健太郎

    情報アクセスシンポジウム2011 2011年9月

  224. Toward Plan Recognition in Discourse Using Large-Scale Lexical Resources.

    Naoya Inoue

    The Seventeenth Annual Meeting of the Association for Natural Language Processing. 2011年3月7日

  225. 言論マップ生成技術の現状と課題

    水野 淳太, Eric Nichols, 渡邉 陽太郎, 村上 浩司, 松吉 俊, 大木 環美, 松本 裕治

    言語処理学会第17回年次大会 2011年3月7日

  226. Toward Evidence Search.

    Eric Nichols, Junta Mizuno, Yotaro Watanabe

    Proceedings of The Seventeenth Annual Meeting of the Association for Natural Language Processing 2011年3月7日

  227. 大規模Web情報分析のための分析対象ページの段階的選択.

    赤峯享, 加藤義清, 川田拓也, レオン末松豊インティ, 河原大輔, 黒橋禎夫

    言語処理学会第17回年次大会 2011年3月7日

  228. 情報分析システムWISDOMのユーザ評価とその分析.

    川田拓也, 赤峯享, 河原大輔, 加藤義清, 黒橋禎夫, 木俵豊

    言語処理学会第17回年次大会 2011年3月7日

  229. Web文書の時間・論理関係分析に基づく情報信頼性判断支援システムの開発と実証実験.

    岡嶋穣, 河合剛巨, 中澤聡, 村上浩司, 松吉俊, 水野淳太, エリック・ニコルズ, 渡邉陽太郎, 渋木英潔, 中野正寛, 宮崎林太郎, 石下円香, 森辰則

    言語処理学会第17回年次大会. 2011年3月7日

  230. 自然言語処理の新展開とウェブ情報編集の未来像

    東北大学大学院国際文化研究科第11回「言語・脳・認知」コロキアム 2010年12月

  231. 文間の弱い対立関係の認識

    大木 環美, 村上 浩司, 松吉 俊, 水野 淳太, 乾 健太郎, 松本 裕治

    情報処理学会研究報告 第199回自然言語処理研究会 2010年11月

  232. 大規模言語/世界知識に基づく次世代Web情報分析

    電子情報通信学会コンピュテーション研究会 2010年10月

  233. コーパスへの意味的注釈の重層的付与

    英語コーパス学会第36回大会シンポジウム「コーパスにおける言語的注釈と自然言語処理技術」 2010年10月

  234. コミュニケーションの条件と言語情報編集技術

    東北大学情報科学研究科第4回総合科学を考えるセミナー 2010年10月

  235. Toward Evidence Search

    Eric Nichols, Kentaro Inui

    NLP若手の会 第5回シンポジウム 2010年9月

  236. 文間関係認識における局所構造アライメントの有効性

    水野 淳太, 増田 祥子, 村上 浩司, 渡邉 陽太郎, 乾 健太郎

    NLP若手の会 第5回シンポジウム 2010年9月

  237. 文章に潜在する書き手の心的状態の推定に向けて

    井之上直也, 乾健太郎

    NLP若手の会 第5回シンポジウム 2010年9月

  238. 帰属文書数に基づくWebページ情報発信者の専門性分析

    加藤義清, 乾健太郎, 黒橋禎夫

    情報処理学会情報基礎とアクセス技術研究会 2010年7月

  239. 大規模知識に基づく言語情報処理と次世代ウェブ情報編集の可能性

    北海道大学大学院情報科学研究科GCOE講演会 2010年6月

  240. 大規模言語/世界知識に基づくWeb情報編集の高度化.

    電子情報通信学会思考と言語研究会 2010年5月

  241. 文間関係認識のための局所構造アライメント

    水野淳太, 後藤隼人, 渡邉陽太郎, 村上浩司, 松本裕治

    第196回自然言語処理研究会 2010年5月

  242. 日本語文章における直接照応及び間接照応の統合的解析

    井之上直也, 飯田龍, 松本裕治

    情報処理学会第72回全国大会 2010年3月

  243. 文間意味的関係認識による言論マップ生成

    村上浩司, 水野淳太, 後藤隼人, 大木環美, 松吉俊, 松本裕治

    言語処理学会第16回年次大会 2010年3月

  244. 文間の限定関係認識:課題設計および分析と予備実験

    大木環美, 村上浩司, 水野淳太, 増田祥子, 松本裕治

    言語処理学会第16回年次大会 2010年3月

  245. 文間関係認識のための構造的アライメント

    後藤隼人, 水野淳太, 村上浩司, 松本裕治

    言語処理学会第16回年次大会 2010年3月

  246. モダリティ、真偽情報、価値情報を統合した拡張モダリティ解析

    江口萌, 松吉俊, 佐尾ちとせ, 松本裕治

    言語処理学会第16回年次大会 2010年3月

  247. Web ページの情報発信構成の同定

    加藤義清, 河原大輔, 黒橋禎夫

    言語処理学会第16回年次大会 2010年3月

  248. Web 情報の俯瞰的把握のための主要・対比・対立文の抽出と集約.

    河原大輔, 黒橋禎夫

    言語処理学会第16回年次大会 2010年3月

  249. Web情報分析のための大規模Webページの収集・選択・検索

    赤峯享, 加藤義清, 河原大輔, レオン末松豊インティ, 新里圭司, 黒橋禎夫, 木俵豊

    言語処理学会第16回年次大会 2010年3月

  250. 機械翻訳研究の課題と新方式の展望

    乾健太郎

    機械翻訳シンポジウム 2009年10月

  251. 次世代Web情報編集のための言語情報処理技術

    乾健太郎

    ニフティ技術講演会 2009年10月

  252. 日本語書き言葉コーパスへの重層的意味情報付与 〜照応・共参照,述語項構造,モダリティ,談話関係〜,

    飯田龍

    科研費特定研究「日本語コーパス」平成21年度全体会議 2009年9月

  253. 日本語文章の事象に対する判断情報アノテーション

    江口萌, 松吉俊, 佐尾ちとせ, 松本裕治

    情報処理学会研究報告, 自然言語処理研究会 2009年9月

  254. Webページの大規模収集・検索基盤

    赤峯享, 加藤義清, 河原大輔, 新里圭司, 黒橋禎夫, 木俵豊

    情報処理学会研究会 2009年7月

  255. 隠れ変数を持つ条件付き確率場による依存構造木の評価極性分類

    中川哲治, 黒橋禎夫

    情報処理学会研究報告, 自然言語処理研究会 2009年7月

  256. 複数文書から抽出した言明間の意味的関係の整理と関係付与

    村上浩司, 増田祥子, 松吉俊, Eric Nichols, 松本裕治

    情報処理学会研究報告, 自然言語処理・言語理解とコミュニケーション合同研究会 2009年7月

  257. コンピュータが言葉の意味を理解できると何が変わるか? --検索エンジンを超えるもの--

    乾健太郎

    NTTコミュニケーション科学基礎研究所未来想論 2009年6月

  258. 言語情報編集と意味処理基盤

    乾健太郎

    大阪市立大学大学院創造都市研究科ワークショップ 2009年6月

  259. SIGVerse:身体的認知・運動を伴う社会エージェントのためのシミュレータープラットフォーム

    稲邑哲也, 柴田智広, 瀬名秀明, 橋本敬, 川合伸幸, 宮下敬宏, 櫻井圭記, 清水正宏, 大武美保子, 細田耕, 梅田聡, 乾健太郎

    人工知能学会全国大会 2009年6月

  260. 社会的インタラクションを実装可能なシミュレータープラットフォーム 〜SIGVerse: 社会的知能発生シミュレータ〜.

    稲邑哲也, 柴田智広, 瀬名秀明, 橋本敬, 川合伸幸, 宮下敬宏, 櫻井圭記, 清水正宏, 大武美保子, 細田耕, 梅田聡, 乾健太郎

    日本機械学会ロボティクス・メカトロニクス講演会 2009年5月

  261. 指定指示・代行指示を区別した指示連体詞の照応解析

    井之上直也, 飯田龍, 松本裕治

    言語処理学会第15回年次大会 2009年3月

  262. 言明間の意味的関係の体系化とコーパス構築

    村上浩司, 増田祥子, 松吉俊, 松本裕治

    言語処理学会第15回年次大会 2009年3月

  263. みんなの経験:ブログから抽出したイベントおよびセンチメントのDB化

    阿部修也, 江口萌, 隅田飛鳥, 大崎梓

    言語処理学会第15回年次大会 2009年3月

  264. Constructing a Scientific Blog Corpus for Information Credibility Analysis

    Eric Nichols, Koji Murakami, Kentaro Inui, Yuji Matsumoto

    言語処理学会第15回年次大会 2009年3月

  265. ウェブニュースを利用した雑談対話システム

    水野淳太, 松本裕治

    人工知能学会言語・音声理解と対話処理研究会 2009年3月

  266. 雑談対話のための評価表現を利用する相槌.

    清水友裕, 松本裕治

    人工知能学会言語・音声理解と対話処理研究会 2009年3月

  267. 根拠情報抽出の課題設計と予備実験

    飯田龍, 松本裕治

    言語処理学会第15回年次大会 2009年3月

  268. 日本語文のモダリティ・極性情報を捉えるために.

    佐尾ちとせ, 江口萌, 松吉俊

    言語処理学会第15回年次大会 2009年3月

︎全件表示 ︎最初の5件までを表示

産業財産権 5

  1. 危険予測装置、運転支援システム

    井之上直也, 奥野英一, 安部克則, 近藤敏之, 栗谷康隆, 乾健太郎

    産業財産権の種類: 特許権

  2. 管理装置、管理方法及び管理プログラム

    小阪尚子, 上江洲愛, 爰川知宏, 乾健太郎, 山口健史

    産業財産権の種類: 特許権

  3. 危機対応支援装置及び危機対応支援方法

    上江洲愛, 小阪尚子, 爰川知宏, 乾健太郎, 山口健史

    産業財産権の種類: 特許権

  4. 危機対応マネジメント装置

    上江洲愛, 小阪尚子, 爰川知宏, 種茂文之, 乾健太郎, 山口健史

    産業財産権の種類: 特許権

  5. 対話処理装置

    乾健太郎, 井之上直也, 佐藤祥多, 渡部生聖, 樋口佐和

    産業財産権の種類: 特許権

共同研究・競争的資金等の研究課題 22

  1. 深い論述理解の計算モデリングと論述学習支援への応用

    乾 健太郎, 井之上 直也, 中川 智皓, HEINZERLING BENJAMIN, 吉川 将司

    提供機関:Japan Society for the Promotion of Science

    制度名:Grants-in-Aid for Scientific Research

    研究種目:Grant-in-Aid for Scientific Research (A)

    研究機関:Tohoku University

    2022年4月1日 ~ 2027年3月31日

  2. 実効性ある偽情報・誤情報対応策としてのファクトチェック・システムの構築

    瀬川 至朗, 乾 健太郎, 関谷 直也, 牧野 恵美

    提供機関:Japan Society for the Promotion of Science

    制度名:Grants-in-Aid for Scientific Research

    研究種目:Grant-in-Aid for Scientific Research (B)

    研究機関:Waseda University

    2019年4月1日 ~ 2023年3月31日

    詳細を見る 詳細を閉じる

    実効性あるファクトチェック・システムの要件を探るため、ファクトチェックの有効性とその限界、そしてファクトチェッカーの選択基準についての研究を進めた。 岡本(2021)は、2018年9月の沖縄県知事選挙のときに行われたファクトチェックについて、関連するツイートデータの分析を実施した。琉球新報社がスペクティ社の協力を得て収集した沖縄県知事選関連ツイートを瀬川が研究目的で提供を受けており、そのツイートデータを分析対象とした。得られた結果から、ファクトチェック記事が有効に機能する条件として、「ファクトチェック主体の知名度や拡散力」など3点が示唆された。また、疑義言説の支持アカウント群と否定アカウント群(ファクトチェック記事支持を含む)はほとんど交じり合っておらず、政治的な分断状況が見て取れた。ファクトチェック情報と対象言説のツイッター上の反応をより正確に把握するには、リツイートを含むデータの分析が求められる。研究グループは沖縄県知事選に関連し、その分析に取り組んでいる。 池(2021)は、近年、ファクトチェック活動が増加している日本と台湾のファクトチェッカーが対象言説をどう選んでいるのか、その選択基準などについて、選挙関連ファクトチェックの分析・調査を実施した。ファクトチェック活動が市民の信頼を得るには「非党派性と公正性」が重要だと指摘されている。その観点から、自民党政権が長い日本と二大政党制の台湾におけるファクトチェックの特徴と課題を考察した。 このほか、新型コロナ感染症の偽・誤情報とファクトチェックの問題に関連して、関谷らは市民の情報受容についてのアンケート調査を実施した。 また、ファクトチェックへのAI導入の世界的潮流を学ぶため、専門家であるスコット・ヘール氏(MEEDANリサーチディレクター)を招いてオンライン研究会(一般公開)を開催し、AI導入の具体的な取り組みを学び、議論した。

  3. 実世界知識と言語知識に基づく処理過程の説明性を備えた文章変換計算機構

    鈴木 潤, 松林 優一郎, 乾 健太郎, 赤間 怜奈

    提供機関:Japan Society for the Promotion of Science

    制度名:Grants-in-Aid for Scientific Research

    研究種目:Grant-in-Aid for Scientific Research (B)

    研究機関:Tohoku University

    2019年4月1日 ~ 2022年3月31日

    詳細を見る 詳細を閉じる

    人間と同等レベルの文章生成計算機構の実現は,人工知能および自然言語処理研究における最重要かつ最難関な未解決課題の一つである.現在の深層ニューラルネットに基づく方法は流暢な文章を生成できるようになった反面,誤生成問題という新たな課題が指摘されている. 本研究では,この誤生成問題の解消および発生メカニズムの解明を目的として,事例を用いて誤生成の要因を判断する方法やその根拠として利用するために自然言語解析の高精度化などいくつかの重要な研究を行い問題解決に向けた研究成果を創出した.

  4. 論理推論・機械学習・物理計算の融合によって「行間を読む」談話解析モデル 競争的資金

    乾 健太郎

    2015年 ~ 2018年

  5. 21世紀市民のための高次リテラシーと批判的思考力のアセスメントと育成

    楠見 孝, 子安 増生, 道田 泰司, MANALO Emmanuel, 林 創, 平山 るみ, 信原 幸弘, 坂上 雅道, 原 塑, 三浦 麻子, 小倉 加奈代, 乾 健太郎, 田中 優子, 小口 峰樹, 沖林 洋平

    提供機関:Japan Society for the Promotion of Science

    制度名:Grants-in-Aid for Scientific Research

    研究種目:Grant-in-Aid for Scientific Research (A)

    研究機関:Kyoto University

    2011年4月1日 ~ 2016年3月31日

    詳細を見る 詳細を閉じる

    本研究は,課題1-1「市民リテラシーと批判的思考のアセスメント」では市民リテラシーを支える批判的思考態度を検討し,評価ツールを開発した。課題1-2「批判的思考育成のための教育プログラム作成と授業実践」では,学習者間相互作用を重視した教育実践を高校・大学において行い,効果を分析した。課題2「神経科学リテラシーと科学コミュニケーション」では,哲学と神経生理学に基づいて推論と情動を検討した。さらに市民主体の科学コミュニケーション活動を検討した。課題3「ネットリテラシーと情報信頼性評価」では,放射能リスクに関する情報源信頼性評価とリテラシーの関連を調査によって解明し,情報信頼性判断支援技術を開発した。

  6. ネット上の医療・健康情報の健全な利活用を可能にするヘルスリテラシー支援環境 競争的資金

    乾 健太郎

    2011年 ~ 2014年

    詳細を見る 詳細を閉じる

    本研究では,ネットユーザの情報信憑性判断を支援し,その能力を育成するヘルスリテラシー支援環境の研究開発を行うもので,本年度は次の3つの研究項目に取り組んだ.(A)相対言論空間生成(乾,岡崎,渡邉)・医療・健康情報に関連するWeb文書を数百万ページの規模で収集したドメインコーパスを用意し,これと医療関連知識ベースであるライフサイエンス辞書を用いて,種々の言語解析モジュールを医療・健康情報ドメインにチューニングした.・ドメインコーパスの一部に文脈情報や言明間関係を人手で付与した注釈付きコーパスを作成した.これにより,文脈処理や言明間関係の仕様を医療・健康情報ドメインに適応させる.得られた注釈付きコーパスは,今後モデルの訓練・評価用データとして用いる予定である.・現在の文間関係認識モジュールを医療・健康情報ドメインにチューニングする.(B)ヘルスリテラシー支援環境(乾,楠見)・ドメインチューニングした文間関係認識モジュールをベースに想定し,ヘルスリテラシー支援環境の基本設計を行った.(C)ヘルスリテラシーの支援・教育方法(楠見)・一般市民および患者・家族を対象とする大規模なネット行動調査を行うための種々の準備を行った.

  7. 経験マイニング技術の高度化と実用化 競争的資金

    乾 健太郎

    2009年 ~ 2010年

    詳細を見る 詳細を閉じる

    ウェブには、社会的な大事件から個人の小さな経験まで、あるいは専門家の評論から一般ユーザの意見まで、さまざまな事実や意見の膨大な情報が日々刻々蓄積されている。こうした情報の山は、使い方によっては人類共通の「知の宝庫」となる可能性を秘めている。本研究では、ウェブ文書中に言葉で書かれた事実情報や意見情報をその意味内容に基づいて整理する言語情報編集技術の開発に取り組んだ。事実や意見の情報を整理する方法には、個々の情報をさまざまな観点から分類し、大きな表のような構造をつくる方法と、個々の情報の間の関係をとらえ、ネットワーク型の構造をつくる方法が考えられる。分類によって整理する方法については、ウェブ文書に書かれた個人の経験情報を<トピック,経験主,事態タイプ,事実性>という一般性の高い4種類の意味的な軸で分類し、巨大な経験データベースを自動生成する経験マイニング技術を世界で初めて開発した。「視力が戻る(=ポジティブな出来事)」のような幅広い事態表現を分類し、さらに事態の成立、不成立、願望などの事実性を区別して経験情報を収集する。これにより、たとえば「あるサービスの利用に伴うトラブル」や「ある商品に関心を持ちながらまだ買っていない人」など、複雑な検索も可能になる。一方、関係によって整理する方法については、別々の文書に書かれた情報の間の隠れた根拠関係を自動的に検出する技術を開発した。前述の経...

  8. 書き言葉コーパスの自動アノテーションの研究 競争的資金

    2006年 ~ 2009年

    詳細を見る 詳細を閉じる

    コーパスに対する自動アノテーションツールの開発:係り受け解析において最適の係り先を比較によって決定する手法の提案,述語項構造解析のための手法の事態名詞への拡張を行い,システムの開発の過程で作成した共参照および項構造情報が付与されたコーパスを公開した。 コーパス管理ツールの開発:形態素・係り受け解析コーパス管理ツールに対していくつかの機能拡張を行った。特に、係り受け構造の表示機能の充実、および、高頻度の連続・非連続のパターンのマイニング機能などを開発した。汎用アノテーションツールについてデータ構造の設計の詳細化といくつかの機能拡張を行った。さらに、Webベースのコーパス検索ツールの開発を行った。 大規模な固有表現辞書の構築:Wikipediaの構造情報を利用し、自動的に固有表現を抽出する手法を提案し、実際に大規模な固有表現抽出を行った。抽出した大規模固有表現辞書とコーパスからの統計的手法を組み合わせることにより、精度向上を図った。白書コアデータ対する固有表現タグ付け作業を行った。 談話構造アノテーションツールの開発:文関数、共参照、項構造を記述するためセマンティックエディタの拡張を行った.一般化された木構造表示ユーザインタフェスの実装を行い、談話構造や述語に対する統語的構成素の記述や文章の埋め込み構造、意味構造を編集する機能をこのユーザインタフェースにプラグインする形で実装した。

  9. 経験マイニング : Web文書からの個人の経験の抽出と分類 競争的資金

    乾 健太郎

    2007年 ~ 2008年

    詳細を見る 詳細を閉じる

    商品やサービスなど, 指定されたトピックに関連する個人の経験の記述をWeb文書集合から収集し, 述語項構造に基づく表現形式に構造化するとともに, 事態タイプ(ポジティブ/ネガティブな出来事・状態, 入手・利用等の行為など)や事実性情報(当該事態の時間情報とそれに対する話者態度)といった意味情報を解析する経験マイニングを開発した. 20年度の具体的成果は次の4点である.(1)評価極性知識獲得の大規模実験 : 事態タイプのうち, とくに「遅刻する, 炎症が治まる, 錆が出る」など, 評価極性を持つ出来事に関する知識の獲得に注力し, 大規模なWeb文書コーパスからこれを獲得する実験を行った. その結果, コーパスのサイズを大きくすると, 獲得できる知識の精度, カバレッジともに劇的に向上に, 最終的に1.6億文のコーパスから75以上のカバレッジを85以上の精度で獲得できることが確かめられた.(2)事実性解析モデルの洗練 : 事実性解析については, 2007年度の成果をベースに, 事実性タグ体系の見直しと訓練データの拡張を行った. また, 文中で隣接する事態表現の事実性の間に依存関係があることに着目し, これをFactorial CRFでモデル化することによって解析精度を向上させることができた.(3)公開デモサイト「みんなの経験」の開発 : 以上の成果を利用し, 文書集合から実際に経...

  10. 語構成を考慮した多言語の語彙および用語解析システムの研究開発 競争的資金

    2006年 ~ 2008年

    詳細を見る 詳細を閉じる

    次の3項目に分けて研究を実施した. (1)中国語および日本語の合成語の語構成の整理と分類:中国語と日本語の合成語を複合語と派生語に分類し、語構成に関する分類を行なった。特に、中国語の3文字からなる合成語について、構造の記述を網羅的に行い、自動分類に関する実験を行った。 (2)言語解析用辞書の語彙項目の拡充:形態素解析器「茶筌」用の辞書の拡張を目指して、大規模テキストからの未知語の自動抽出に関する検討を行った。また、中国語と日本語の新規の複合語の抽出と登録を行った。中国語について、合討約13万語の辞書を構築した。 (3)複合表現・専門用語の自動解析手法の開発:特定の分野に頻出する用語の抽出と意味分類を目指し、Web上の大規模なデータからの自動抽出に関する研究を行った.特に、インターネット上で構築が進んでいるWikipediaを対象とし、そのDOM構造を利用して,用語の意味分類を行なう手法を提案し、固有表現辞書の自動構築を行った。

  11. 語彙意味論に基づく言い換え計算機構の工学的実現と言い換え知識獲得への適用 競争的資金

    乾 健太郎

    提供機関:Ministry of Education, Culture, Sports, Science and Technology

    制度名:Grants-in-Aid for Scientific Research(基盤研究(B))

    研究種目:Grant-in-Aid for Scientific Research (B)

    研究機関:Nara Institute of Science and Technology

    2005年 ~ 2007年

    詳細を見る 詳細を閉じる

    本研究の目的は,動詞などの述語とその項からなる述語項構造を事態表現の基本単位と仮定し,語彙意味論等の言語学的知見をふまえて述語項構造間の同義・含意関係に関する基本知識を設計整備することであった.これに対し,(1)動詞語釈文の構造化,(2)語彙概念構造(LCS)に基づく事態上位オントロジーの構築,(3)コーパスからの事態間関係獲得,の互いに相補的な3つのアプローチを提案し,以下の成果を得た.(1)動詞語釈文の構造化 国語辞典の語釈文から述語項構造間の基本的な意味関係を収集する研究に取り組んだ.例えば,動詞「倒す」の語釈文「立っている物に力を加え傾け、横にする」からは,「XがYを倒す→XがYを横にする(上位下位関係)」の他,「→XがYに力を加える(行為-手段関係)」,多様な意味関係が収集できる.実際に岩波国語辞典第5版の収録動詞(11469語)から8種類の意味関係を合わせて約3万5千件収集した.(2)LCSに音尽く事態上位オントロジーの構築 LCSにおける動詞意味分析の枠組みに基づき,高頻度動詞約4千語,7千語義について5階層からなる意味分類を行った.最下層は約千クラスに分類されており,例えば,「所属・客体変化」のクラスには「配属する,取り立てる,引き抜く,立てる,招く」などが属し,これらはすべて同じ項構造を持つ.これにより,(1)の語釈文の構造化だけでは捉えられない基本概念間の...

  12. 円滑な情報伝達を支援する言語規格と言語変換技術 競争的資金

    2004年 ~ 2007年

    詳細を見る 詳細を閉じる

    本研究の主な成果は次のとおりである。 1. 日本語の中核となる語彙を列挙した日本語基本語彙JC2.2を編纂した。この語彙表は、211の機能表現と177の慣用表現を含む、5814語(表現)を収録している。 2. 機能表現341語とその出現形16,711件を階層的に整理した、日本語機能表現辞書を編纂した。この辞書を用いて、機能表現を言い換えるシステムを実現した。 3. 「現代語複合辞用例集」に準拠した日本語複合辞用例データベースMUST1を編纂した。このデータベースを利用した機械学習により、テキスト中の機能表現を検出するシステムを実現した。 4. 小中高の文部科学省検定教科書と大学教養課程の教科書からサンプルテキストを抽出した教科書コーパス(サンプルテキスト数1478、総文字数約100万字)を編纂した。この教科書コーパスを規準テキストとして、日本語テキストの難易度を推定するシステムを実現した。 5. 述語句に対して、多数の同義表現を生成・提示するフレーザル・シソーラスを実現した。 6. 応用指向の言い換え・言語変換として、「サ変動詞+名詞」の複合名詞への言い換え、理解補助を目指した動詞句の言い換え、「新幹線要約」のための文末の言い換えを実現した。また、要約の新しい手法として、要約事例を用例として利用したニュース記事要約を提案・実現した。 7. 上記成果のうち、JC2.2, MUS...

  13. 言語研究のためのコーパスの作成と利用に関する研究 競争的資金

    2003年 ~ 2005年

    詳細を見る 詳細を閉じる

    言語解析ツールに関する研究として,これまで開発してきた日本語の形態素解析,係り受け解析システムを中国語の品詞タグ付けおよび単語係り受け解析システムとして実装した. また,辞書の整備については,中国語の未知語解析法の整備を行い,大量のテキストデータから未知語を抽出する実験を行い,10万語を越える大規模な中国語辞書を作成した.また,日本語の複合表現の辞書への登録に関する基準を設定して複合語辞書の整備を行い,複合語の内部構造を記述した日本語辞書を作成した.英語についても,複合表現の品詞推定についての実験を行い,複合表現が字義的に使われているかどうかを自動判定する手法について高い精度を与える成果を得た. コーパス管理ツールについては,タグ付き言語データと辞書項目の関連付けと関係データベースへの格納のフォーマットの詳細化を行い,これまで開発してきたタグ付きコーパス管理システムに実装した.コーパス中の単語分かち書きおよび品詞タグ付けに関する誤りを修正するためのインタフェースの設計と実装を行った.係り受け解析が行われた文のデータベースへの格納法を決定し,また,係り受け構造を表示するためのインタフェースの実装を行った.さらに,そのインタフェースを用いて,文節のまとめあげと係り受けに関するタグ付け誤りを修正する機能を実装した. 構築したタグ付きコーパス管理システムの諸機能の使い易さと正しさの検...

  14. 文書集合からの事態関係知識の自動獲得と深い言語理解への応用 競争的資金

    乾 健太郎

    2003年 ~ 2005年

    詳細を見る 詳細を閉じる

    本研究は,大規模な事態関係知識を文書集合から抽出し,深い言語理解の実現に利用する技術の確立を目的とするものである.計画立案当初は3年間で(a)事態間の因果関係に関する知識(以下,因果関係知識.Casuse(<地震が起こる>,<津波が起こる>),Effect(<X_<具体物>を干す>,<X_<具体物>が乾く>)など),および(b)事態の望ましさに関する単項関係知識(Positive(<健康である>),Negative(<津波が起こる>など)の2種類の知識を対象とする計画だったが,研究代表者が平成15年度2月より日本学術振興会在外研究のため渡英することになったので,(a)のタイプの因果関係知識の獲得に絞って研究を実施した.まず,因果関係に立つ2つの出来事がそれぞれ行為であるか事態であるかに基づき,cause,effect,precond(ition),meansの4つの因果関係を定義した.接続標識「ため」を含む複文からこれら4つの因果関係への自動分類実験を行い,cause,precond,meansの各関係について,80%の再現率で95%以上の分類精度を達成した.また,effect関係については,30%の再現率で90%の分類精度,を達成した.さらに,新聞記事1年分の文書集合に本手法を適用することにより,およそ27,000件を超える因果知識が獲得できる見積りを得た.研究費は主として...

  15. Semantic Webの高度化のためのオントロジーの自動生成に関する研究 競争的資金

    2003年 ~ 2004年

    詳細を見る 詳細を閉じる

    Semantic Webは、オントロジーに基づくコンテンツの意味内容記述に基づくグローバルな知識ベースであり、近年盛んに研究活動が行われている。そのベースとなるオントロジーを作成するために、辞典のような自然言語で記述・体系化された知識源の、人間と機械の共同作業による意味構造化を目指す。そこで、昨年度に引き続き、用語辞典(認知科学辞典)コンテンツをXML(Extensible Markup Language)によって構造化し、用語の定義文に対して、言語構造や語彙情報に基づく言語的アノテーションの付与を行った。これは、言語的アノテーションに関する研究代表者らの研究成果であるアノテーション作成支援システムを用いることによって、半自動的に行った。 今年度は、昨年度の実験結果を踏まえて、辞典項目への言語的アノテーションによって構築される用語のネットワーク構造の修正・補完を行った。たとえば、ネットワーク構造の構成要素に欠落があり、関係が正しく導かれない場合は、XML構造化されたコーパスを用いて、欠けている部分の補完を試みた。論文のカテゴリー分け、未知語の認定などには機械学習の手法を用いた。人間は、コーパスから得られるネットワーク構造を吟味し、用語のネットワーク構造に付け加える作業を行った。同一の用語が異なる分野あるいは異なる文脈で使われる現象を、機械学習の手法を用いて発見し、用語を細分化...

  16. 既存の言語資源からの大規模語彙的言い換え知識の自動抽出 競争的資金

    乾 健太郎

    2001年 ~ 2002年

    詳細を見る 詳細を閉じる

    語彙的言い換えは,入力文中の語句を同じ意味を持つ別の語句に置換する作業である.ただし,実際には,文脈に無関係に置換可能な言い換え対はほとんど存在しないため,与えられた言い換え対が置換可能かどうかを所与の文脈に照らして評価する機構が必要になる.本研究では,語彙的言い換えの方法として,同概念語への言い換えと国語辞典の語釈文への言い換えの二つを検討し,実験によって評価機構の有効性を経験的に評価した.初年度は,言い換え事例約1000件を分析し,言い換え後の文に対する適格性評価の項目と修正処理を分類・整理した.得られた知見は以下の通りである.(a)もっとも頻度が高かったのは活用の誤りだが,これの解決は難しくない.(b)次に多かったのは,動詞と格要素の共起が不適格なケースである.第2年度に解決を試みた.(c) (c)と同様に多かったのは,言い換え前の語が多義であり,誤った言い換え対を選択したケースである.しかし,これらの約半数は(b)と同様の方法で解決できることがわかった.(d)その他の問題は上の3種類に比べると頻度がかなり少なかった.以上の観察に基づいて,第2年度は,主として上記(b)の問題にとりくんだ.具体的には,20年分の新聞記事コーパスから<名詞,助詞,動詞>の共起事例を大量に収集し,それを正例の訓練データとした.また,評価対象とする名詞と動詞を頻度に基づいて制限し,この制限を満...

  17. 統計情報と言語的制約情報の融合に関する研究 競争的資金

    2000年 ~ 2002年

    詳細を見る 詳細を閉じる

    電子化された言語データの増加に伴い、統計的言話処理が盛んに研究されているが、多くは表面的な言語処理が対象であり、詳細な意味解析など深い言語処理を行うには適切でない。一方で、主辞駆動句構造文法に代表される制約に基づく文法では、個々の単語に詳細な言語情報を記述し、単語固有の文法現象を説明することができるが、制約に合致する文の間の解釈の優先度の違いを説明できないこと、制約に少しでも違反する文の解析を行うことができないなどの欠点がある。 本研究では、これら両者の欠点を補うため、高精度の単語係り受け解析を統計情報に基づく頑健な方法で実現し、得られた統語構造を利用して、制約に基づく文法による詳細な解析を行うことを目的とした。その際、制約違反を緩和するために、生成語彙理論で導入されたタイプ強制や共構成という演算を実装するためのソフトウェアシステムとインタフェースを構築した。 係り受け解析については、サポートベクターマシンという機械学習を用い、上昇型に決定的に解析を行うアルゴリズムを提案し、日本語および英語の解析システムを構築した。制約に基づく文法として、主辞駆動句構造文法に基づく日本語文法を引き続き開発するとともに、タイプ強制と共構成という高次の生成演算を行うメカニズムを実装した。また、統計解析および文法情報を受け取って、文解析を行い、その過程を表示するためのユーザインタフェースを構築した。

  18. 自己増殖型言語知識ベース構築技術に関する研究 競争的資金

    1997年 ~ 1999年

    詳細を見る 詳細を閉じる

    本研究課題では,自然言語処理に有用な言語知識ベースのうち,特に形態素情報・構文情報が付与されたコーパスに着目し,これを自己増殖的に構築する手法に関する研究を行った.その手法の概要は以下の通りである.まず,大量のテキストの形態素・構文解析を行い,コーパスに付与すべき形態素情報・構文情報を得る.次に,得られた形態素情報・構文情報から,形態素・構文解析に必要な知識や解析精度を向上させるための知識として,接続表と確立一般化LRモデルを学習する.接続表とは,品詞対の隣接可能性に関する知識であり,コーパス内に一度でも隣接して現われた品詞対は隣接可能,それ以外の品詞対は隣接不可能として,品詞タグ付きコーパスから自動的に獲得することができる.一方,確率一般化LRモデルは,解析結果の統計的な意味での正しさの尺度を構文木に与える確率モデルであり,そのパラメタはコーパスに付与された構文情報(構文木)から容易に学習することができる.さらに,得られた知識をもとに再びテキストの形態素・構文解析を行い,コーパスに付与するべき新たな形態素情報・構文情報を得る.このことを繰り返し行うことにより,形態素情報・構文情報が付与されたコーパスを自動的に作成する.実験の結果,本研究課題で提案する手法は,既存の形態素情報・構文情報が付与されたコーパスが存在するとき,その規模を拡大する際に,すなわち形態素・構文情報が付与さ...

  19. 関連する複数のテキストを対象とした要約生成技術に関する研究 競争的資金

    1997年 ~ 1998年

    詳細を見る 詳細を閉じる

    従来の情報検索システムでは,検索対象となる文書と検索要求とを重み付きのキ-ワ-ドベクトルで表現し,余弦を求めることで関連性を判定するベクトル空間モデルを用いることが多い.通常,ベクトルの各要素の重みの計算には出現回数を基にした統計量を用いるが,このような表層的な情報のみを利用した手法では,関連性が低い文書も検索してしまうという問題が生ずる.今年度は,従来の手法によって検索された文書集合を対象とし,その中から自然言語処理技術を用いて不適切な文書を排除する手法について研究をおこなった.具体的には,文書中の動詞,格要素(名詞,助詞)からなる格フレ-ムに注目し,格フレ-ムと検索要求との関連性を求めることによって文書と検索要求との関連性を判定する.格フレ-ムを利用することで,文書の構造的な情報を扱うことが可能となり,キ-ワ-ドのみの集合に比べて精密に文書の内容を把握することが可能となる.初期検索によって検索された文書集合から,格フレ-ムと検索要求との関連性の計算する.関連性の判定には,格フレ-ムがどの程度検索要求に対して特徴的であるかを示す重要度という尺度を用いる.格フレ-ムに対して直接重要度をめることは難しいため,動詞,格要素それぞれに重要度を求め,それらを用いて格フレ-ム重要度を求める.また類似,包含関係にある格フレ-ム間では,類似性に応じて相互に重要度を修正することにより,特徴的...

  20. タスク指向日本語対話システムにおける漸進的発話の生成方法に関する研究 競争的資金

    乾 健太郎

    1997年 ~ 1998年

    詳細を見る 詳細を閉じる

    人間どうしの対話を観察すると,複雑な情報をより単純な情報に分割し,少しずつ互いの理解を深めていくように工夫していることがわかる.本研究では,このような対話の漸進性に着目し,対話システム上で漸進的発話を生成する計算機構の構築をはかった.研究は以下の手順で進めた.まず,対話コーパスを分析し,対話に現れる漸進的発話についてその表現上の特徴と意味・意図の分類を行った.対話コーパスには,電話と地図による道案内模擬対話データを用いた.この模擬対話では,発話権の移動を制限していない.分析の結果,個々の断片的発話について,従来の対話研究に見られる発話意図よりも粒度の細かい微細な発話意図が認められた.そこで我々は,道案内対話という領域に限定して,微細な発話意図を手作業により網羅的に収集し,分類した.この結果,個々の断片的発話の意図を複数の微細な発話意図の組み合わせで表現できることがわかった.つぎに,この分析にもとづき,微細な発話意図と断片的発話の対応関係を単一化システムCUF上に記述することによって,漸進的発話生成システムを試作した.本システムは,微細な発話意図を組み合わせて適当な大きさの発話プランを作る処理と,発話プランを表層の言語表現に変換する処理を擬似並列的に行う.後者の処理のための言語知識はシステミック文法に基づく記述方法によって実装してある.生成実験の結果,道案内に領域を限れば,ある...

  21. 種々の統計情報を統合した統計的自然言語解析に関する研究 競争的資金

    1997年 ~ 1997年

    詳細を見る 詳細を閉じる

    本研究では,構文解析の曖昧性解消を行うために,構文的な統計情報と語彙的な統計情報を統合する手法を提案する.我々が提案する統合的確率言語モデルは,構文的優先度などの構文的な統計情報を反映する構文モデルと,単語の出現頻度や単語の共起関係などの語彙的な統計情報を反映する語彙モデルの2つの下位モデルから成る.この統合的確率言語モデルは,構文的な統計情報と語彙的な統計情報を同時に学習する 去の多くのモデルと異なり,両者を個別に学習する点に特徴がある.構文的な統計情報と語彙的な統計情報を独立に取り扱うことにより,それぞれの統計情報を異なる言語資源から独立に学習することができるだけでなく,それぞれの統計情報が曖昧性解消においてどのような効果を果たすのかを容易に分析することができる.この統合的確率言語モデルを評価するために,日本語文の文節の係り受け解析を行った.構文モデルを用いたときの文節の正解率は73.38$%となり,ベ-スラインに比べて11.70$%向上した.また,構文モデルと語彙モデルを組み合わせることにより,文節の正解率はさらに10.96$%向上し84.34$%となった.この結果,本研究で提案する枠組において,語彙的な統計情報は構文的な統計情報と同程度に曖昧性解消に貢献することを確認した.なお,本研究の成果は下記国際会議で報告されている。 Kiyoaki Shirai,Kentaro...

  22. タスク指向日本語対話システムにおけるあいづちの理解と生成に関する研究 競争的資金

    乾 健太郎

    1996年 ~ 1996年

    詳細を見る 詳細を閉じる

    対話における情報の送り手は,受け手にとって新しい情報の伝達を言わば小出しに行い,受け手の反応に応じて情報を付加していく.本研究では,このようにして作られる送り手の発話を漸進的発話と呼び,その生成と理解のための計算モデルの構築をはかった.具体的には以下の点を明らかにした.1.漸進的発話の結果生じる個々の断片的発話について、その言語的・文脈的特徴と発話意図との対応関係を対話コーパスから分析し,断片的発話の意図理解のためのヒューリスティクスとしてまとめた.2.対話コーパスの分析をさらに進め,個々の発話断片を形成する発話意図が微細な発話意図の組み合わせとして表現できることを明らかにし、これを計算機上で扱うための内部表現を設計した.3.微細な発話意図の集合を入力とし,これを必要に応じて組み合わせ,漸進的発話の発話プランを生成する単一化ベースのアルゴリズムを開発し,発話プランニングモジュールとして実装した.4.発話プランを体系機能文法に基づいて表出する発話表出モジュールとして実装した.5.実装したシステムを用いて,対話コーパスに現れる漸進的発話を網羅的に生成する実験を行い,アルゴリズムの評価,および生成に必要な言語資源の拡充を行った.研究成果は平成8年12月に人工知能学会言語・音声理解と対話処理研究会(SIG-J-9601-8)で発表し,また平成9年3月には言語処理学会第2回年次大会で発...

︎全件表示 ︎最初の5件までを表示

社会貢献活動 2

  1. 高校大学連携講座

    2011年10月3日 ~

    詳細を見る 詳細を閉じる

    千葉県立千葉高校に出張し,高校生を相手に言語情報処理の先進的技術に関する講義を行った.

  2. 高校大学連携講座

    2010年7月21日 ~

    詳細を見る 詳細を閉じる

    県下佐沼高校に出張し,高校生を相手に言語情報処理の先進的技術に関する講義を行った.

その他 9

  1. 知識の構造化に関する研究

    詳細を見る 詳細を閉じる

    知識の構造化に関する研究

  2. 高齢者・家族間コミュニケーション時の高齢者応答自動生成と、高齢者の心の状態推論のためのQOL推論知識構築に関する検討

    詳細を見る 詳細を閉じる

    高齢者・家族間コミュニケーション時の高齢者応答自動生成と、高齢者の心の状態推論のためのQOL推論知識構築に関する検討

  3. テキストからの知識抽出技術と知識の活用技術の研究

    詳細を見る 詳細を閉じる

    テキストからの知識抽出技術と知識の活用技術の研究

  4. 自然言語処理を活用した危機対応ログに基づくリスク予測・レコメンド手法に関わる研究開発

    詳細を見る 詳細を閉じる

    自然言語処理を活用した危機対応ログに基づくリスク予測・レコメンド手法に関わる研究開発

  5. ロボットコミュニケーションのための常識知識獲得技術の研究

    詳細を見る 詳細を閉じる

    ロボットコミュニケーションのための常識知識獲得技術の研究

  6. 自然言語処理技術・知識情報処理技術の産業応用に関する学術指導

    詳細を見る 詳細を閉じる

    自然言語処理技術・知識情報処理技術の産業応用に関する学術指導

  7. 静止画・動画からのシーン理解とイベント予測

    詳細を見る 詳細を閉じる

    静止画・動画からのシーン理解とイベント予測

  8. 言語情報間の論理関係の自動推論による言論空間の可視化

    詳細を見る 詳細を閉じる

    Web上の情報の信憑性の判断を支援することを目的として,Web上の言語情報間の論理関係を自動推論し,隠れた言論空間を可視化する技術を研究開発する.

  9. 先端的知識獲得の研究

    詳細を見る 詳細を閉じる

    高度な言語意味解析により、Webなどの大量言語情報から世界知識を獲得する研究

︎全件表示 ︎最初の5件までを表示