TOHOKU UNIVERSITY Researchers - Keisuke Sakaguchi

Details of the Researcher

Home

日本語 English

Keisuke Sakaguchi

Section

Graduate School of Information Sciences

Job title

Associate Professor

Degree

Ｐｈ．Ｄ．（Johns Hopkins University）
修士（工学）（奈良先端科学技術大学院大学）
Ｍ．Ａ．（University of Essex）

researchmap

https://researchmap.jp/sakaguchi-keisuke

J-GLOBAL ID

202101000370288104

Research History 8

2022/07 - Present

Tohoku University　Associate Professor
2018/08 - 2022/06

Allen Institute for Artificial Intelligence　Research Scientist
2013/08 - 2018/08

Johns Hopkins University　Research Assistant
2022/09 - Present

RIKEN　Center for Advanced Intelligence Project (AIP)　Visiting researcher
2017/05 - 2017/08

IBM T.J. Watson Research Center　Research Intern
2014/05 - 2014/08

Educational Testing Service (ETS)　Research Intern
2012/08 - 2012/11

Microsoft Research Asia　Research Intern
2011/08 - 2011/09

Yahoo! Japan 研究所　インターン

Show all Show first 5

Education 4

Johns Hopkins University　Department of Computer Science　Ph.D in Computer Science

2013/08 - 2018/08
Nara Institute of Science and Technology

2011/04 - 2013/03
University of Essex　Department of Language and Linguistics　MA in Psycholinguistics and Neurolinguistics

2005/10 - 2006/09
Waseda University　School of Letters, Arts and Sciences I　Philosophy

2001/04 - 2005/03

Committee Memberships 14

AAAI Conference on Artificial Intelligence　Program Committee

2020 - Present
Conference on Computational Natural Language Learning (CoNLL)　Program Committee

2019 - Present
International Joint Conferences on Artificial Intelligence (IJCAI)　Senior Program Committee

2019 - Present
Transactions of the Association for Computational Linguistics (TACL)　Program Committee

2019 - Present
Workshop on Noisy User-generated Text (W-NUT)　Program Committee

2018 - Present
Annual Meeting of the North American Chapter of the ACL (NAACL)　Program Committee

2018 - Present
Annual Meeting of the Association for Computational Linguistics (ACL)　Program Committee

2018 - Present
Computational Linguistics Journal　Program Committee

2018 - Present
International Joint Conference on Natural Language Processing (IJCNLP)　Program Committee

2017 - Present
Conference on Empirical Methods on Natural Language Processing (EMNLP)　Program Committee

2015 - Present
Workshop on Innovative Use of NLP for Building Educational Applications (BEA)　Program Committee

2013 - Present
International Conference on Spoken Language Translation (IWSLT)　Program Committee

2020 - 2020
International Conference on Computational Linguistics (COLING)　Program Committee

2014 - 2020
Methods for Optimizing and Evaluating Neural Language Generation (NeuralGen)　Program Committee

2019 - 2019

Show all ︎Show first 5

Research Interests 2

Artificial Intelligence
Natural Language Processing

Research Areas 1

Informatics / Intelligent informatics / Natural Language Processing

Awards 6

言語処理学会第30回年次大会優秀賞

2024/03　言語処理学会　長文生成の多面的評価:人手評価と自動評価の向上を目指して
Best Paper Award

2023　AACL-IJCNLP Student Research Workshop
言語処理学会第28回年次大会優秀賞

2022/03　言語処理学会　論述リビジョンのためのメタ評価基盤
AAAI 2020 Outstanding Paper Award (Best Paper)

2020/02　AAAI (Association for the Advancement of Artificial Intelligence)　WinoGrande: An Adversarial Winograd Schema Challenge at Scale
ACL 2017 Outstanding Paper Award

2017/08　Association for Computational Linguistics (ACL)　Error-repair Dependency Parsing for Ungrammatical Texts
言語処理学会論文賞

2017　言語処理学会　Phrase Structure Annotation and Parsing for Learner English

Show all ︎Show 5

Papers 52

An Empirical Investigation of Machines' Capabilities for Moral Judgment with the Delphi Experiment Peer-reviewed

Liwei Jiang, Jena D. Hwang, Chandra Bhagavatula, Ronan Le Bras, Jenny Liang, Jesse Dodge, Keisuke Sakaguchi, Maxwell Forbes, Jon Borchard, Saadia Gabriel, Yulia Tsvetkov, Oren Etzioni, Maarten Sap, Regina Rini, Yejin Choi

Nature Machine Intelligence　2025/01
RealTime QA: What's the Answer Right Now? Peer-reviewed

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

NeurIPS　2023
I2D2: Inductive Knowledge Distillation with NeuroLogic and Self-Imitation. Peer-reviewed

Chandra Bhagavatula, Jena D. Hwang, Doug Downey, Ronan Le Bras 0001, Ximing Lu, Lianhui Qin, Keisuke Sakaguchi, Swabha Swayamdipta, Peter West, Yejin Choi 0001

ACL (1)　9614-9630　2023

DOI： 10.18653/v1/2023.acl-long.535 　
WinoGrande Invited Peer-reviewed

Keisuke Sakaguchi, Ronan Le Bras, Chandra Bhagavatula, Yejin Choi

Communications of the ACM　64　(9)　99-106　2021/09
Publisher: Association for Computing Machinery (ACM)
DOI： 10.1145/3474381 　

ISSN： 0001-0782

eISSN： 1557-7317

More details Close

Commonsense reasoning remains a major challenge in AI, and yet, recent progresses on benchmarks may seem to suggest otherwise. In particular, the recent neural language models have reported above 90% accuracy on the Winograd Schema Challenge (WSC), a commonsense benchmark originally designed to be unsolvable for statistical models that rely simply on word associations. This raises an important question---whether these models have truly acquired robust commonsense capabilities or they rely on spurious biases in the dataset that lead to an overestimation of the true capabilities of machine commonsense. To investigate this question, we introduce WinoGrande, a large-scale dataset of 44k problems, inspired by the original WSC, but adjusted to improve both the scale and the hardness of the dataset. The key steps of the dataset construction consist of (1) large-scale crowdsourcing, followed by (2) systematic bias reduction using a novel AFLITE algorithm that generalizes human-detectable <italic>word associations</italic> to machine-detectable <italic>embedding associations.</italic> Our experiments demonstrate that state-of-the-art models achieve considerably lower accuracy (59.4%-79.1%) on WINOGRANDE compared to humans (94%), confirming that the high performance on the original WSC was inflated by spurious biases in the dataset. Furthermore, we report new state-of-the-art results on five related benchmarks with emphasis on their dual implications. On the one hand, they demonstrate the effectiveness of WINOGRANDE when used as a resource for transfer learning. On the other hand, the high performance on all these benchmarks suggests the extent to which spurious biases are prevalent in all such datasets, which motivates further research on algorithmic bias reduction.
Abductive Commonsense Reasoning. Peer-reviewed

Chandra Bhagavatula, Ronan Le Bras, Chaitanya Malaviya, Keisuke Sakaguchi, Ari Holtzman, Hannah Rashkin, Doug Downey, Wen-tau Yih, Yejin Choi

ICLR　2020
Publisher: OpenReview.net
Towards Automated Document Revision: Grammatical Error Correction, Fluency Edits, and Beyond.

Masato Mita, Keisuke Sakaguchi, Masato Hagiwara, Tomoya Mizumoto, Jun Suzuki 0001, Kentaro Inui

Proceedings of the 19th Workshop on Innovative Use of NLP for Building Educational Applications(BEA)　251-265　2024
Publisher: Association for Computational Linguistics
J-UniMorph: Japanese Morphological Annotation through the Universal Feature Schema

Kosuke Matsuzaki, Masaya Taniguchi, Kentaro Inui, Keisuke Sakaguchi

Proceedings of the 21st SIGMORPHON workshop on Computational Research in Phonetics, Phonology, and Morphology　7-19　2024
Publisher: Association for Computational Linguistics
DOI： 10.18653/v1/2024.sigmorphon-1.2 　
A Multimodal Dialogue System to Lead Consensus Building with Emotion-Displaying

Shinnosuke Nozue, Yuto Nakano, Shoji Moriya, Tomoki Ariyama, Kazuma Kokuta, Suchun Xie, Kai Sato, Shusaku Sone, Ryohei Kamei, Reina Akama, Yuichiroh Matsubayashi, Keisuke Sakaguchi

Proceedings of the 25th Annual Meeting of the Special Interest Group on Discourse and Dialogue　669-673　2024
Publisher: Association for Computational Linguistics
DOI： 10.18653/v1/2024.sigdial-1.57 　
How Well Do Vision Models Encode Diagram Attributes?

Haruto Yoshida, Keito Kudo, Yoichi Aoki, Ryota Tanaka, Itsumi Saito, Keisuke Sakaguchi, Kentaro Inui

ACL (Student Research Workshop)　564-575　2024
First Heuristic Then Rational: Dynamic Use of Heuristics in Language Model Reasoning. Peer-reviewed

Yoichi Aoki, Keito Kudo, Tatsuki Kuribayashi, Shusaku Sone, Masaya Taniguchi, Keisuke Sakaguchi, Kentaro Inui

EMNLP　2024

DOI： 10.48550/arXiv.2406.16078 　
The Curse of Popularity: Popular Entities have Catastrophic Side Effects when Deleting Knowledge from Language Models.

Ryosuke Takahashi, Go Kamoda, Benjamin Heinzerling, Keisuke Sakaguchi, Kentaro Inui

CoRR　abs/2406.06032　2024

DOI： 10.48550/arXiv.2406.06032 　
ACORN: Aspect-wise Commonsense Reasoning Explanation Evaluation. Peer-reviewed

Ana Brassard, Benjamin Heinzerling, Keito Kudo, Keisuke Sakaguchi, Kentaro Inui

COLM (CONFERENCE ON LANGUAGE MODELING)　2024

DOI： 10.48550/arXiv.2405.04818 　
Hagi bot: A Multimodal Dialogue System for Smooth Discussion with Human-like Behavior and Dialogue State Tracking Using LLM

NAKANO Yuto, NOZUE Shinnosuke, KOKUTA Kazuma, ARIYAMA Tomoki, SATO Kai, SONE Shusaku, KAMEI Ryohei, XIE Suchun, NARITA Fuka, MORIYA Shoji, AKAMA Reina, MATSUBAYASHI Yuichiroh, SAKAGUCHI Keisuke

JSAI Technical Report, SIG-SLUD　99　102-107　2023/12/04
Publisher: The Japanese Society for Artificial Intelligence
DOI： 10.11517/jsaislud.99.0_102 　

ISSN： 0918-5682

eISSN： 2436-4576

More details Close

This paper describes "hagi bot," a system submitted to the Sixth Dialogue System Live Competition. It is a task-oriented multi-modal dialogue system that integrates a response generation module and an avatar control module. Within the response generation module, GPT-4 is used to generate responses along with emotion and action labels considering the dialogue history and the topics to be discussed. Specifically, by monitoring the dialogue state through slot filling and continuously changing prompts based on the situation, the module is able to achieve a natural dialogue progression. In the avatar control module, voice (pitch, volume, and speaking speed), facial expressions, and gestures are regulated based on predefined rules designed in reference to models such as Russell's Circumplex Model of Affect, corresponding to content of speech together with emotion and action labels. This approach enables human-like natural behavior. The combination of these two modules achieves generation of responses appropriate to the situation and natural behaviors based on the content of utterances and emotions. This system won the first place in the preliminary round.
Test-time Augmentation for Factual Probing.

Go Kamoda, Benjamin Heinzerling, Keisuke Sakaguchi, Kentaro Inui

CoRR　abs/2310.17121　2023

DOI： 10.48550/arXiv.2310.17121 　
Test-time Augmentation for Factual Probing. Peer-reviewed

Go Kamoda, Benjamin Heinzerling, Keisuke Sakaguchi, Kentaro Inui

EMNLP (Findings)　3650-3661　2023
PlaSma: Making Small Language Models Better Procedural Knowledge Models for (Counterfactual) Planning. Peer-reviewed

Faeze Brahman, Chandra Bhagavatula, Valentina Pyatkin, Jena D. Hwang, Xiang Lorraine Li, Hirona J. Arai, Soumya Sanyal 0001, Keisuke Sakaguchi, Xiang Ren 0001, Yejin Choi 0001

ICLR (The Twelfth International Conference on Learning Representations.)　abs/2305.19472　2023

DOI： 10.48550/arXiv.2305.19472 　
Evaluating GPT-4 and ChatGPT on Japanese Medical Licensing Examinations.

Jungo Kasai, Yuhei Kasai, Keisuke Sakaguchi, Yutaro Yamada, Dragomir Radev

CoRR　abs/2303.18027　2023

DOI： 10.48550/arXiv.2303.18027 　
Causal schema induction for knowledge discovery.

Michael Regan, Jena D. Hwang, Keisuke Sakaguchi, James Pustejovsky

CoRR　abs/2303.15381　2023

DOI： 10.48550/arXiv.2303.15381 　
An Analysis of GPT-3's Performance in Grammatical Error Correction.

Steven Coyne, Keisuke Sakaguchi

CoRR　abs/2303.14342　2023

DOI： 10.48550/arXiv.2303.14342 　
ELQA: A Corpus of Metalinguistic Questions and Answers about English. Peer-reviewed

Shabnam Behzad, Keisuke Sakaguchi, Nathan Schneider 0001, Amir Zeldes

ACL (1)　2031-2047　2023

DOI： 10.18653/v1/2023.acl-long.113 　
Do Deep Neural Networks Capture Compositionality in Arithmetic Reasoning? Peer-reviewed

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

Proceedings of the 17th Conference of the European Chapter of the Association for Computational Linguistics(EACL)　1343-1354　2023
Publisher: Association for Computational Linguistics
DOI： 10.18653/v1/2023.eacl-main.98 　
Empirical Investigation of Neural Symbolic Reasoning Strategies. Peer-reviewed

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

Findings of the Association for Computational Linguistics: EACL 2023　1124-1132　2023
Publisher: Association for Computational Linguistics
DOI： 10.18653/v1/2023.findings-eacl.86 　
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

DOI： 10.48550/arXiv.2205.11484 　
Bidimensional Leaderboards: Generate and Evaluate Language Hand in Hand. Peer-reviewed

Jungo Kasai, Keisuke Sakaguchi, Ronan Le Bras, Lavinia Dunagan, Jacob Morrison, Alexander R. Fabbri, Yejin Choi, Noah A. Smith

NAACL-HLT　3540-3557　2022
Transparent Human Evaluation for Image Captioning. Peer-reviewed

Jungo Kasai, Keisuke Sakaguchi, Lavinia Dunagan, Jacob Morrison, Ronan Le Bras, Yejin Choi, Noah Smith

NAACL-HLT　3464-3478　2022
proScript: Partially Ordered Scripts Generation via Pre-trained Language Models. Peer-reviewed

Keisuke Sakaguchi, Chandra Bhagavatula, Ronan Le Bras, Niket Tandon, Peter Clark, Yejin Choi

EMNLP　abs/2104.08251　2021
(Comet-) Atomic 2020: On Symbolic and Neural Commonsense Knowledge Graphs. Peer-reviewed

Jena D. Hwang, Chandra Bhagavatula, Ronan Le Bras, Jeff Da, Keisuke Sakaguchi, Antoine Bosselut, Yejin Choi

AAAI　6384-6392　2021
Publisher: AAAI Press
The Universal Decompositional Semantics Dataset and Decomp Toolkit. Peer-reviewed

Aaron Steven White, Elias Stengel-Eskin, Siddharth Vashishtha, Venkata Subrahmanyan Govindarajan, Dee Ann Reisinger, Tim Vieira, Keisuke Sakaguchi, Sheng Zhang 0012, Francis Ferraro, Rachel Rudinger, Kyle Rawlins, Benjamin Van Durme

LREC　5698-5707　2020
Publisher: European Language Resources Association
A Dataset for Tracking Entities in Open Domain Procedural Text. Peer-reviewed

Niket Tandon, Keisuke Sakaguchi, Bhavana Dalvi, Dheeraj Rajagopal, Peter Clark, Michal Guerquin, Kyle Richardson 0001, Eduard H. Hovy

EMNLP　6408-6417　2020
Publisher: Association for Computational Linguistics
DOI： 10.18653/v1/2020.emnlp-main.520 　
Uncertain Natural Language Inference. Peer-reviewed

Tongfei Chen, Zhengping Jiang, Adam Poliak, Keisuke Sakaguchi, Benjamin Van Durme

ACL　8772-8779　2020
Publisher: Association for Computational Linguistics
DOI： 10.18653/v1/2020.acl-main.774 　
WinoGrande: An Adversarial Winograd Schema Challenge at Scale. Peer-reviewed

Keisuke Sakaguchi, Ronan Le Bras, Chandra Bhagavatula, Yejin Choi

AAAI　8732-8740　2020
Publisher: AAAI Press
WIQA: A dataset for "What if..." reasoning over procedural text. Peer-reviewed

Niket Tandon, Bhavana Dalvi, Keisuke Sakaguchi, Peter Clark, Antoine Bosselut

EMNLP　6075-6084　2019
Publisher: Association for Computational Linguistics
DOI： 10.18653/v1/D19-1629 　
Efficient Online Scalar Annotation with Bounded Support. Peer-reviewed

Keisuke Sakaguchi, Benjamin Van Durme

ACL　208-218　2018
Publisher: Association for Computational Linguistics
DOI： 10.18653/v1/P18-1020 　
Grammatical Error Correction with Neural Reinforcement Learning. Peer-reviewed

Keisuke Sakaguchi, Matt Post, Benjamin Van Durme

IJCNLP　366-372　2017
Publisher: Asian Federation of Natural Language Processing
JFLEG: A Fluency Corpus and Benchmark for Grammatical Error Correction. Peer-reviewed

Courtney Napoles, Keisuke Sakaguchi, Joel R. Tetreault

EACL　229-234　2017
Publisher: Association for Computational Linguistics
DOI： 10.18653/v1/e17-2037 　
GEC into the future: Where are we going and how do we get there? Peer-reviewed

Keisuke Sakaguchi, Courtney Napoles, Joel R. Tetreault

BEA　180-187　2017
Publisher: Association for Computational Linguistics
DOI： 10.18653/v1/w17-5019 　
Error-repair Dependency Parsing for Ungrammatical Texts. Peer-reviewed

Keisuke Sakaguchi, Matt Post, Benjamin Van Durme

ACL　189-195　2017
Publisher: Association for Computational Linguistics
DOI： 10.18653/v1/P17-2030 　
Robsut Wrod Reocginiton via Semi-Character Recurrent Neural Network. Peer-reviewed

Keisuke Sakaguchi, Kevin Duh, Matt Post, Benjamin Van Durme

AAAI　3281-3287　2017
Publisher: AAAI Press
Reassessing the Goals of Grammatical Error Correction: Fluency Instead of Grammaticality Peer-reviewed

Keisuke Sakaguchi, Courtney Napoles, Matt Post, Joel Tetreault

Transactions of the Association for Computational Linguistics　4　169-182　2016/12
Publisher: MIT Press - Journals
DOI： 10.1162/tacl_a_00091 　

eISSN： 2307-387X

More details Close

The field of grammatical error correction (GEC) has grown substantially in recent years, with research directed at both evaluation metrics and improved system performance against those metrics. One unvisited assumption, however, is the reliance of GEC evaluation on error-coded corpora, which contain specific labeled corrections. We examine current practices and show that GEC’s reliance on such corpora unnaturally constrains annotation and automatic evaluation, resulting in (a) sentences that do not sound acceptable to native speakers and (b) system rankings that do not correlate with human judgments. In light of this, we propose an alternate approach that jettisons costly error coding in favor of unannotated, whole-sentence rewrites. We compare the performance of existing metrics over different gold-standard annotations, and show that automatic evaluation with our new annotation scheme has very strong correlation with expert rankings (ρ = 0.82). As a result, we advocate for a fundamental and necessary shift in the goal of GEC, from correcting small, labeled error types, to producing text that has native fluency.
GLEU Without Tuning.

Courtney Napoles, Keisuke Sakaguchi, Matt Post, Joel R. Tetreault

CoRR　abs/1605.02592　2016
There's No Comparison: Reference-less Evaluation Metrics in Grammatical Error Correction. Peer-reviewed

Courtney Napoles, Keisuke Sakaguchi, Joel R. Tetreault

EMNLP　2109-2115　2016
Publisher: The Association for Computational Linguistics
DOI： 10.18653/v1/d16-1228 　
Universal Decompositional Semantics on Universal Dependencies. Peer-reviewed

Aaron Steven White, Drew Reisinger, Keisuke Sakaguchi, Tim Vieira, Sheng Zhang 0012, Rachel Rudinger, Kyle Rawlins, Benjamin Van Durme

EMNLP　1713-1723　2016
Publisher: The Association for Computational Linguistics
DOI： 10.18653/v1/d16-1177 　
Phrase Structure Annotation and Parsing for Learner English. Peer-reviewed

Ryo Nagata, Keisuke Sakaguchi

ACL　2016
Publisher: The Association for Computer Linguistics
DOI： 10.18653/v1/p16-1173 　
Effective Feature Integration for Automated Short Answer Scoring. Peer-reviewed

Keisuke Sakaguchi, Michael Heilman, Nitin Madnani

NAACL　1049-1054　2015
Publisher: The Association for Computational Linguistics
DOI： 10.3115/v1/n15-1111 　
Ground Truth for Grammaticality Correction Metrics. Peer-reviewed

Courtney Napoles, Keisuke Sakaguchi, Matt Post, Joel R. Tetreault

ACL　588-593　2015
Publisher: The Association for Computer Linguistics
DOI： 10.3115/v1/p15-2097 　
Efficient Elicitation of Annotations for Human Evaluation of Machine Translation. Peer-reviewed

Keisuke Sakaguchi, Matt Post, Benjamin Van Durme

WMT　1-11　2014
Publisher: The Association for Computer Linguistics
DOI： 10.3115/v1/w14-3301 　
Construction of English MWE Dictionary and its Application to POS Tagging. Peer-reviewed

Yutaro Shigeto, Ai Azuma, Sorami Hisamoto, Shuhei Kondo, Tomoya Kouse, Keisuke Sakaguchi, Akifumi Yoshimoto, Frances Yung, Yuji Matsumoto 0001

MWE　139-144　2013
Publisher: The Association for Computer Linguistics
NAIST at 2013 CoNLL Grammatical Error Correction Shared Task.

Ippei Yoshimoto, Tomoya Kose, Kensuke Mitsuzawa, Keisuke Sakaguchi, Tomoya Mizumoto, Yuta Hayashibe, Mamoru Komachi, Yuji Matsumoto 0001

Proceedings of the Seventeenth Conference on Computational Natural Language Learning: Shared Task　26-33　2013
Publisher: ACL
NAIST at the NLI 2013 Shared Task.

Tomoya Mizumoto, Yuta Hayashibe, Keisuke Sakaguchi, Mamoru Komachi, Yuji Matsumoto 0001

Proceedings of the Eighth Workshop on Innovative Use of NLP for Building Educational Applications(BEA@NAACL-HLT)　134-139　2013
Publisher: The Association for Computer Linguistics
Discriminative Approach to Fill-in-the-Blank Quiz Generation for Language Learners. Peer-reviewed

Keisuke Sakaguchi, Yuki Arase, Mamoru Komachi

ACL　238-242　2013
Publisher: The Association for Computer Linguistics
Joint English Spelling Error Correction and POS Tagging for Language Learners Writing. Peer-reviewed

Keisuke Sakaguchi, Tomoya Mizumoto, Mamoru Komachi, Yuji Matsumoto 0001

COLING　2357-2374　2012
Publisher: Indian Institute of Technology Bombay
NAIST at the HOO 2012 Shared Task.

Keisuke Sakaguchi, Yuta Hayashibe, Shuhei Kondo, Lis Kanashiro, Tomoya Mizumoto, Mamoru Komachi, Yuji Matsumoto 0001

Proceedings of the Seventh Workshop on Building Educational Applications Using NLP(BEA@NAACL-HLT)　281-288　2012
Publisher: The Association for Computer Linguistics

Show all ︎Show first 5

Misc. 12

自然画像で学習された画像埋め込みにダイアグラムを特徴づける情報は含まれているか?

吉田遥音, 工藤慧音, 工藤慧音, 青木洋一, 青木洋一, 田中涼太, 田中涼太, 斉藤いつみ, 坂口慶祐, 坂口慶祐, 乾健太郎, 乾健太郎, 乾健太郎

言語処理学会年次大会発表論文集(Web)　30th　2024

ISSN： 2188-4420
Towards grammatically-informed feedback comments

GALVAN-SOSA Diana, COYNE Steven, SAKAGUCHI Keisuke, INUI Kentaro, GALVAN-SOSA Diana, COYNE Steven, SAKAGUCHI Keisuke, INUI Kentaro

言語処理学会年次大会発表論文集(Web)　29th　2023

ISSN： 2188-4420
因果的プロンプトによるNLIの敵対的ロバスト性の強化

KAVUMBA Pride, KAVUMBA Pride, BRASSARD Ana, BRASSARD Ana, HEINZERLING Benjamin, HEINZERLING Benjamin, 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

ISSN： 2188-4420
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

ISSN： 2188-4420
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

ISSN： 2188-4420
大規模言語モデルにおける暗黙の推論生成能力の評価

根岸直生, 坂口慶祐, 乾健太郎, 坂口慶祐, 乾健太郎

情報科学技術フォーラム講演論文集　22nd　2023
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

ISSN： 2188-4420
英語学習者コーパスのための句構造アノテーション

永田亮, 坂口慶祐

言語処理学会年次大会発表論文集(Web)　21st　2015

ISSN： 2188-4420
Construction of English Multiword Dictionary and its Application to POS Tagging

Yutaro Shigeto, Ai Azuma, Shuhei Kondo, Ryuta Kitaura, Keisuke Sakaguchi, Tomoya Kose, Sorami Hisamoto, Akifumi Yoshimoto, Frances Yung, Yuji Matsumoto

IPSJ SIG Notes　2012　(7)　1-6　2012/11/15
Publisher: Information Processing Society of Japan (IPSJ)

More details Close

Many previous studies have proposed various methods for English Part-Of-Speech (POS) tagging, and these methods have been shown to give high accuracy. However POS tagging in these previous methods are only on each word, and they cannot tag on multiword expressions consisting of multiple lexical items, such as collocations. In this research we constructed a multiword expression dictionary, and we used variable length CRF to conduct POS tagging which takes into consideration the multiword expressions. We, then, evaluated the tagging accuracy and the multiword recognition accuracy. In result, the tagger with the multiword dictionary achieved higher accuracy compared with the tagger which does not consider multiword expressions.
Joint Learning of English Spelling Error Correction and POS Tagging

Keisuke Sakaguchi, Tomoya Mizumoto, Mamoru Komachi, Yuji Matsumoto

IPSJ SIG Notes　2012　(8)　1-7　2012/05/03
Publisher: Information Processing Society of Japan (IPSJ)

More details Close

Automated grammatical error detection and correction tasks for the second language (L2) learners writing of English have become more important in recent years. L2 writing contains not only grammatical errors but also other types of errors such as misspelling and punctuation errors. These `ungrammatical' errors often disturb part-of-speech (POS) tagging and dependency parsing, resulting in an obstacle for grammatical error detection and correction tasks. Spelling error correction and POS tagging have been studied independently but in recent years joint learning of related tasks has been successful in improving NLP pipeline processing. In this paper, we propose a joint learning approach to English spelling error correction and POS tagging. The experimental result shows that the proposed method can correct spelling errors and label POS tags simultaneously for L2 writing as well or better than applying each method independently.
英作文誤り訂正における複数の手法の利用に関する考察

水本智也, 林部祐太, 坂口慶祐, 小町守, 松本裕治

情報処理学会研究報告(CD-ROM)　2012　(3)　2012

ISSN： 2186-2583
オークション検索クリックスルーログからの属性値抽出

水本智也, 坂口慶祐, 小町守, 内海慶, 河野洋志, 前澤敏之, 佐藤敏紀

言語処理学会年次大会発表論文集　18th (CD-ROM)　2012

Show all ︎Show first 5

Presentations 23

RealTime QA: What's the Answer Right Now? Invited

Keisuke Sakaguchi

2023 BrainLink X-Lab Day, Korean Federation of Science and Technology Societies (KOFST), Republic of Korea　2023/12/03
ChatGPTにできること・できないこと (得意なこと・不得意なこと) Invited

坂口慶祐

言語処理学会言語処理技術セミナー　2023/10/30
自然言語処理 Invited

坂口慶祐

大阪大学基礎工学部・生物工学特論　2023/10/20
人間とAIの対話：ChatGPTがもたらす変革と課題 Invited

坂口慶祐

第11回岡山大学AI研究会　2023/09/13
人間とAIの対話:ChatGPTがもたらす変革と課題 Invited

坂口慶祐

学術変革領域 A「尊厳学の確立:尊厳概念に基づく社会統合の学際的パラダイムの構築に向けて」尊厳学フォーラム第3回 B04 班シンポジウム「生成 AI と尊厳」東京大学，東京　2023/09/10
Twist Decoding: Diverse Generators Guide Each Other Invited

Keisuke Sakaguchi

CWRU-TOHOKU Joint Workshop, Sendai, Japan　2023/08/07
ことばの不思議と人工知能 Invited

坂口慶祐

河合塾知の広場，仙台　2023/07/28
人間とAIの対話: ChatGPTがもたらす変革と課題 Invited

坂口慶祐

秋桜会, 仙台　2023/07/19
GPT4と大規模言語モデルおよびその応用 Invited

坂口慶祐

JTC オンラインセミナー　2023/07/07
Large Language Models: What is happening now? Invited

Keisuke Sakaguchi

Sungkyunkwan University, Republic of Korea (online)　2023/04/05
Large Language Models: What will happen next? Invited

坂口慶祐

産業技術総合研究所人工知能研究センター，東京　2023/01/23
Large Language Models: What will happen next? Invited

Keisuke Sakaguchi

2022 BrainLink X-Lab Day, Korean Federation of Science and Technology Societies (KOFST), Republic of Korea　2022/10/25
Large Language Models: What will happen next? Invited

Keisuke Sakaguchi

AIST AI Research Center, Japan　2022/09/27
Large Language Models: What will happen next? Invited

坂口慶祐

東京大学松尾研究室，東京　2022/08/23
Robust Text Correction for Grammar and Fluency Invited

Keisuke Sakaguchi

Microsoft Research, Redmond, WA　2018/09/27
Robust Text Correction for Grammar and Fluency Invited

Keisuke Sakaguchi

Allen Institute for Artificial Intelligence, Seattle, WA　2018/03/30
Grammatical Error Correction with Neural Reinforcement Learning Invited

Keisuke Sakaguchi

Grammarly Research, Online　2018/02/22
Robust Text Correction for Grammar and Fluency Invited

Keisuke Sakaguchi

Amazon Research, Boston, MA　2018/02/09
文法および流暢性を考慮した頑健なテキスト誤り訂正 Invited

坂口慶祐

千葉工業大学ステアラボ人工知能セミナー, 東京　2017/12/14
Robust Natural Language Processing Invited

坂口慶祐

NTTコミュニケーション科学研究所, 京都　2017/12/12
Robust Natural Language Processing Invited

坂口慶祐

東北大学乾・岡崎研究室, 仙台　2017/12/08
Robust Natural Language Processing Invited

坂口慶祐

首都大学東京小町研究室, 東京　2017/12/05
Evaluating SMT evaluation Invited

Keisuke Sakaguchi

Rakuten Institute of Technology, NY　2013/11/27

Show all Show first 5

Research Projects 1

現実世界の逐次的環境変化に協調的に適応するマルチモーダル自然言語理解モデル

坂口慶祐

Offer Organization: 日本学術振興会

System: 科学研究費助成事業

Category: 国際共同研究加速基金(帰国発展研究)

Institution: 東北大学

2022/02/18 - 2025/03/31

More details Close

現代社会において、人間とAIによる自然なインタラクションや協働を可能にするためには、ユーザーの文脈を考慮し柔軟に対応できる自然言語処理モデルが必要不可欠である。このようなモデルは、ユーザーのニーズに応じた対話を可能にし、AIとユーザー間のコミュニケーションのをより円滑にすると考えられる。しかし、現時点では「静的な文脈のみを必要とするベンチマーク上での高い精度と、動的な文脈が重要になるアプリケーションでの低い精度とのギャップ」が問題となっている。つまり、一方ではAIは一定の文脈におけるパフォーマンスを向上させる一方で、より広範で複雑な状況に対応する能力にはまだ限界があることが確認されている。その解決策として、本研究課題では現実世界のように常に状況や文脈が変化する環境に対応可能なマルチモーダルモデルを提案する。このモデルは、言語情報だけでなく、視覚情報、聴覚情報を統合的かつ逐次的に学習する能力を持つ。これにより、AIはユーザーの現在の状況をより深く理解し、それに基づいた適切な対応を提供することが可能となる。着任の2022年8月より開始した今年度は、前半に実験環境の構築やシステム開発に向けた関連研究のサーベイ、それらの再現実験、本研究課題において最終的に必要となる、数量推論や知識推論、ユーザーフィードバック生成に関わる深層学習モデルの研究開発を行い、後半以降ではそれらの成果の研究発表（国際会議採択件数3件、国内会議6件、招待講演4件）、および次年度に向けたマルチモーダルモデルのサーベイ、再現実験を開始した。