研究者詳細 - 松原　茂樹

The Power of URLs in Scholarly Papers: Their Role as Metadata Sources for Data Repositories

Wada K., Matsubara S.

Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) 15494 LNCS 巻頁： 85 - 92 2025年

記述言語：英語掲載種別：研究論文（学術雑誌）出版者・発行元：Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

Open science is increasingly important in academia. This paper investigates the citations using URLs, used to reference research resources in scholarly papers, and their potential as metadata sources for data repositories. A dataset of URL citations extracted from scholarly papers in natural language processing is analyzed. We first examine trends in locations and frequency of URL citations over time, then classify the types of research resources and their citation purposes. Furthermore, we assess the usability of URL citations for automatic metadata generation in data repositories. Specifically, we explore how many metadata records can be added and how much the metadata fields can be enriched. Our findings suggest that URL citations offer a promising approach for enhancing metadata quality and availability in data repositories.

DOI： 10.1007/978-981-96-0868-3_6

Capabilities and Challenges of LLMs in Metadata Extraction from Scholarly Papers

その他リンク： https://link.springer.com/chapter/10.1007/978-981-96-0868-3_6

Watanabe Y., Ito K., Matsubara S.

Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) 15493 LNCS 巻頁： 280 - 287 2025年

Research data are cited in scholarly papers, and the construction and use of datasets are mentioned. The descriptions of research data in papers may be used as information for its metadata. In this paper, we focus on large language models (LLM), which have achieved high performance in various natural language processing tasks, and we investigate the ability of LLMs to extract metadata from papers. In the experiment, we analyzed LLMs’ metadata extraction capabilities quantitatively and qualitatively. The results demonstrate that while LLMs can extract metadata from papers extensively, the extraction accuracy is not necessarily high. We confirm that there are challenges in identifying the names of research data and linking information related to the research data.

DOI： 10.1007/978-981-96-0865-2_23

Estimation of Relevance Between Datasets for Enhancing Accessibility of Research Artifacts

その他リンク： https://dl.acm.org/doi/10.1007/978-981-96-0865-2_23

Ito K., Matsubara S.

Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) 15494 LNCS 巻頁： 97 - 103 2025年

Open science is being promoted globally, and researchers are encouraged to publish and share research artifacts, e.g., datasets, tools, software, code, and scholarly papers. To accelerate open science, the accessibility of research artifacts is also important. Previous studies focused on the relevance between scholarly papers to improve their accessibility; however, the relevance between research artifacts other than scholarly papers has not been investigated extensively. In this study, we focus on datasets, and verify the feasibility of estimating the relevance between datasets. Access using the relevance between datasets may be useful when it is difficult to verbalize the characteristics of the desired dataset. We implemented a method to estimate the relevance between datasets that utilizes dataset metadata expressed as text, and the relevance is estimated by using a BERT-based model. Experimental results demonstrated the feasibility of estimating the relevance between datasets. Finally, we discuss the effect of utilizing metadata for each metadata field.

DOI： 10.1007/978-981-96-0868-3_8

その他リンク： https://link.springer.com/chapter/10.1007/978-981-96-0868-3_8

言語処理学会30周年記念事業実施報告 Open Access

井之上直也, 大内啓樹, 河原大輔, 黒橋禎夫, 小町守, 須藤克仁, 永田亮, 松原茂樹, 宮尾祐介

自然言語処理 32 巻 ( 1 ) 頁： 330 - 341 2025年

Negation Scope Conversion for Unifying Negation-Annotated Datasets Open Access

Yoshida Asahi, Kato Yoshihide, Matsubara Shigeki

Journal of Natural Language Processing 32 巻 ( 1 ) 頁： 300 - 329 2025年

記述言語：英語掲載種別：研究論文（学術雑誌）出版者・発行元：The Association for Natural Language Processing

Negation scope resolution is a technique that identifies the part of a sentence affected by the negation cue. The three major corpora used for it, the BioScope corpus, the SFU review corpus, and the Sherlock dataset, have different annotation schemes for negation scope. Due to the different annotations, it is difficult to use the three corpora together in the study of negation scope resolution. To address this issue by merging the corpora into a unified dataset based on a common annotation scheme, we propose a method for automatically converting the scopes of BioScope and SFU to those of Sherlock. We conducted an experiment to evaluate the accuracy of our method using a dataset obtained by manually annotating the negation scopes to a tiny portion of BioScope and SFU, verifying that our method can convert the scopes with high accuracy. In addition, we conducted another experiment to verify the effectiveness of our method from a pragmatic perspective, where we fine-tuned PLM-based negation scope resolution models using the unified dataset obtained by our method. The results demonstrated that the performances of the models increase when fine-tuned on the unified dataset, unlike the simply combined one, which supports the effectiveness of our method.

DOI： 10.5715/jnlp.32.300

Classification of URL Citations on Scholarly Papers Using Intermediate Task Training Open Access

WADA Kazuhiro, TSUNOKAKE Masaya, MATSUBARA Shigeki

IEICE Transactions on Information and Systems advpub 巻 ( 0 ) 2025年

記述言語：英語掲載種別：研究論文（学術雑誌）出版者・発行元：The Institute of Electronics, Information and Communication Engineers

Citations using URL (URL citations) that appear in scholarly papers can be used as an information source for the research resource search engines. In particular, the information about the types of cited resources and reasons for their citation is crucial to describe the resources and their relations in the search services. To obtain this information, previous studies proposed some methods for classifying URL citations. However, their methods trained the model using a simple fine-tuning strategy and exhibited insufficient performance. We propose a classification method using a novel intermediate task. Our method trains the model on our intermediate task of identifying whether sample pairs belong to the same class before being fine-tuned on the target task. In the experiment, our method outperformed previous methods using the simple fine-tuning strategy with higher macro F-scores for different model sizes and architectures. Our analysis results indicate that the model learns the class boundaries of the target task by training our intermediate task. Our intermediate task also demonstrated higher performance and computational efficiency than an alternative intermediate task using triplet loss. Finally, we applied our method to other text classification tasks and confirmed the effectiveness when a simple fine-tuning strategy does not stably work.

DOI： 10.1587/transinf.2024edp7149

Human Performance in Incremental Dependency Parsing: Dependency Structure Annotations and their Analyses 査読有り

Hiroki Unno, Tomohiro Ohno, Koichiro Ito, Shigeki Matsubara

Proceedings of the 37th Pacific Asia Conference on Language, Information and Computation (PACLIC-2024) 2024年12月

研究データをより見つけやすくするためのメタデータ変換と学術機関リポジトリへの登録：太陽地球系物理学分野における実践招待有り

能勢正仁, 新堀淳樹, 三好由純, 堀智昭, 大平司, 岡本麻衣子, 直江千寿子, 我喜屋累, 田中幸恵, 相良毅, 青木学聡, 松原茂樹, 高橋一郎, 林秀和, 山田一成, 南山泰之, 田中良昌, 阿部修司, 上野悟, 今城峻, 齊藤泰雄, 芦北卓也, 堀優子, 清水敏之, 岡村奈々子, 平野かおる

情報の科学と技術 74 巻 ( 11 ) 頁： 487 - 493 2024年11月

記述言語：日本語掲載種別：研究論文（学術雑誌）出版者・発行元：一般社団法人情報科学技術協会

「超高層大気長期変動の全球地上ネットワーク観測・研究（IUGONET）」プロジェクトでは，太陽地球系物理学における地上観測データのメタデータデータベースを構築している。このメタデータデータベースを，幅広い研究コミュニティや一般市民が利用できるように，メタデータスキーマの変換を行い，学術機関リポジトリに登録した。その結果，データベース間の定期的なハーベストにより，学術機関リポジトリデータベース（IRDB），データカタログ横断検索システム，Google Dataset Searchなどにも登録され，研究データの可視性や到達可能性が大きく高まることとなった。また，政策決定機関からの研究メタデータ整備要求に対して，効果的なメタデータマネジメントを実施できた。

DOI： 10.18919/jkg.74.11_487

Utilization of Text Data for Response Timing Detection in Attentive Listening 査読有り

Yu Watanabe, Koichiro Ito, Shigeki Matsubara

Proceedings of Interspeech-2024 頁： 3565 - 3569 2024年9月

On an Intermediate Task for Classifying URL Citations on Scholarly Papers 査読有り

Kazuhiro Wada, Masaya Tsunokake, Shigeki Matsubara

Proceeding of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING2024) 2024年5月

Negation Scope Conversion: Towards a Unified Negation-Annotated Dataset 査読有り

Asahi Yoshida, Yoshihide Kato, Shigeki Matsubara

Proceeding of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING2024) 2024年5月

語りの傾聴において不同意を示す応答の生成査読有り

伊藤滉一朗, 村田匡輝, 大野誠寛, 松原茂樹

自然言語処理 31 巻 ( 1 ) 頁： 212 - 249 2024年3月

担当区分：最終著者記述言語：日本語掲載種別：研究論文（学術雑誌）出版者・発行元：一般社団法人　言語処理学会

コミュニケーションロボットなどの会話エージェントが語りを聴く役割を担うことが期待されている．これらが聴き手として認められるには，語りを傾聴していることを語り手に伝達する機能を備える必要がある．このための明示的な手段は語りに応答することであり，傾聴を示す目的で語りに応答する発話，すなわち傾聴応答の表出が有力である．語りの傾聴では，語り手の発話を受容することが聴き手の基本的な応答方略となる．しかし，語りには，自虐や謙遜などの発話が含まれることがある．この場合，語り手の発話に同意しないことを示す応答，すなわち，不同意応答を確実に表出できることが求められる．本論文では，語りの傾聴において不同意応答を適切に生成することの実現性を示す．そのために，本研究ではまず，時間制約のない環境で語りデータに不同意応答のタイミングと表現をタグ付けする方式を定めた．作成したコーパスを用いて，不同意応答タイミングを網羅的に，不同意応答表現を安定的にタグ付けできることを検証する．続いて，事前学習済みの Transformer ベースのモデルに基づく，不同意応答タイミングの検出手法，及び，不同意応答表現への分類手法を実装し，実験により応答コーパスを用いた不同意応答生成の実現性を検証した．

DOI： 10.5715/jnlp.31.212

Re-Evaluating Syntax-Based Negation Scope Resolution* Open Access

Yoshida, A; Kato, Y; Matsubara, S

IEICE TRANSACTIONS ON INFORMATION AND SYSTEMS E107.D 巻 ( 1 ) 頁： 165 - 168 2024年1月

記述言語：英語掲載種別：研究論文（学術雑誌）出版者・発行元：一般社団法人電子情報通信学会

Negation scope resolution is the process of detecting the negated part of a sentence. Unlike the syntax-based approach employed in previous researches, state-of-the-art methods performed better without the explicit use of syntactic structure. This work revisits the syntax-based approach and re-evaluates the effectiveness of syntactic structure in negation scope resolution. We replace the parser utilized in the prior works with state-of-the-art parsers and modify the syntax-based heuristic rules. The experimental results demonstrate that the simple modifications enhance the performance of the prior syntax-based method to the same level as state-of-the-art end-to-end neural-based methods.

DOI： 10.1587/transinf.2023edl8050

Enhancing Findability and Searchability of Research Data: Metadata Conversion and Registration in Institutional Repositories Open Access

Nosé M., Shinbori A., Miyoshi Y., Hori T., Ohira T., Hashiba J., Naoe C., Gakiya R., Okamoto M., Sagara T., Aoki T., Matsubara S., Takahashi I., Hayashi H., Yamada K., Minamiyama Y., Tanaka Y., Abe S., Ueno S., Imajo S., Saito Y., Ashikita T., Hori Y., Shimizu T., Okamura N., Hirano K., Bargatze L.

Data Science Journal 23 巻 2024年

記述言語：英語掲載種別：研究論文（学術雑誌）出版者・発行元：Data Science Journal

This paper outlines our practice to enhance the findability and searchability of research data through metadata conversion from the Space Physics Archive Search and Extract (SPASE) schema to a more generic schema, the Japan Consortium for Open Access Repository (JPCOAR) schema, and registration of converted metadata in institutional repositories. Traditionally, earth and space science research data have been organized using the SPASE schema. Although the SPASE schema is comprehensive, its usage has been restricted to highly professional databases, limiting broader accessibility and impeding cross-disciplinary research. We discuss a case study at Nagoya University where 284 metadata records were converted from SPASE to JPCOAR, and illustrate the process and benefits of this conversion. This approach significantly improved the visibility and usability of metadata across various platforms like the Institutional Repositories DataBase (IRDB), the Data Catalog Cross-Search System, and Google’s Dataset Search, extending access to a wider range of users beyond the highly professional scientific community. This approach also aligns with national policies in Japan on research data management and simplifies metadata handling for researchers. Future direction includes expanding this conversion and registration model to other universities and institutions by leveraging the ubiquity of the SPASE schema in the earth and space science fields. Our practice may be useful in other research fields. This initiative aims to improve the overall findability of research data, to foster cross-disciplinary collaboration, and to enhance the value of research data itself and of creators and managers of the data.

DOI： 10.5334/dsj-2024-040

Estimating Metadata of Research Artifacts to Enhance their Findability

Ito, K; Matsubara, S

2024 IEEE 20TH INTERNATIONAL CONFERENCE ON E-SCIENCE, E-SCIENCE 2024 頁： 1 - 2 2024年

記述言語：英語掲載種別：研究論文（学術雑誌）出版者・発行元：Proceedings - 2024 IEEE 20th International Conference on e-Science, e-Science 2024

To accelerate open science, it is encouraged to publish and share research artifacts, e.g., scholarly papers, datasets, and code, with their metadata. Although the metadata facilitates the discovery of research artifacts, the metadata is not always complete. In this study, we verify the feasibility of estimating metadata that has not yet been filled (unfilled metadata) using metadata that has been filled (filled-in metadata) for realizing automatic completion of metadata. We implement an estimation method and evaluate its performance through experiments. The method takes filled-in metadata expressed in text as input and estimates unfilled metadata using a BERT-based model. In our experiments, the filled-in metadata was set as the name of the research artifacts, and the unfilled metadata was set as their type. Experimental results showed a high estimation performance.

DOI： 10.1109/e-Science62913.2024.10678684

その他リンク： https://www.computer.org/csdl/proceedings-article/e-science/2024/10678684/20op4XS5gic

Paper Recommendation Using Citation Contexts in Scholarly Documents 査読有り

Tomoki Ikoma, Shigeki Matsubara

Proceedings of the 37th Pacific Asia Conference on Language, Information and Computation (PACLIC 37) 2023年12月

Automatic Insertion of Commas and Linefeeds into Lecture Transcripts Based on Multi-task Learning 査読有り

Zhicheng Fang, Masaki Murata, Shigeki Matsubara:

Proceedings of the 37th Pacific Asia Conference on Language, Information and Computation (PACLIC 37) 2023年12月

On the Use of Language Models for Function Identification of Citations in Scholarly Papers 査読有り

Tomoki Ikoma and Shigeki Matsubara

Proceedings of the 2nd Workshop on Information Extraction from Scientific Publications (WIESP 2023) 2023年11月

Revisiting Syntax-Based Approach in Negation Scope Resolution 査読有り

Asahi Yoshida, Yoshihide Kato and Shigeki Matsubara

Proceedings of the 12th Joint Conference on Lexical and Computational Semantics (*SEM2023) 頁： 18 - 23 2023年7月

ニューラルモデルとタブロー法を用いた自然言語推論査読有り

佐治礼仁, 高尾大樹, 加藤芳秀, 松原茂樹

自然言語処理 30 巻 ( 2 ) 頁： 632 - 663 2023年6月

担当区分：責任著者記述言語：日本語掲載種別：研究論文（学術雑誌）出版者・発行元：一般社団法人　言語処理学会

自然言語推論 (NLI) は，2 つのテキストの間に成り立つ推論的関係を同定するタスクである．近年，ニューラルモデルに基づくアプローチが高い正答率を達成している．しかし，このアプローチに基づくNLIは，判定結果に至る過程や理由を説明する能力を有していないという問題がある．一方，NLI では以前より，記号操作に基づくアプローチが提案されてきた．このアプローチは，推論の論理的な過程を明示でき，推論の根拠を示すことができるものの，語の意味的知識や常識的な知識を十分に備えることは容易でなく，高い推論性能の達成は難しいという問題がある．そこで本論文では，タブロー法とニューラルモデルを組み合わせた手法を提案する．タブロー法は，推論規則の適用に基づく論理式の分解，及び，論理式への真偽値割り当てが存在するか否かの検査から構成される．本手法ではこのうち，真偽値割り当ての検査にニューラルモデルを用いる．なお，タブロー法では通常，論理式を操作対象とするのに対し，本手法では依存構造を対象とする．依存構造を用いることにより，ニューラル NLI モデルをタブロー法アルゴリズムに組み込むことが可能となる．提案手法の論理的整合性を検証するために，手法をモデル理論的に定式化し，その理論的性質を明らかにした．また，SNLI コーパスを用いて推論実験を行い，本手法による推論過程の明示可能性を検証した．

DOI： 10.5715/jnlp.30.632

語りに傾聴を示す応答の表出されやすさの推定査読有り

伊藤滉一朗, 村田匡輝, 大野誠寛, 松原茂樹

電子情報通信学会論文誌A 基礎・境界 J106-A 巻 ( 3 ) 頁： 136 - 145 2023年3月

担当区分：最終著者記述言語：日本語掲載種別：研究論文（学術雑誌）出版者・発行元：電子情報通信学会

コミュニケーションロボットやスマートスピーカーなどの会話エージェントが，人の語りの聴き手を担うことが期待されている．これらが聴き手として認められるには，語りを傾聴していることを語り手に伝える機能を備える必要がある．このための明示的な手段は語りに応答することであり，ジェスチャや発話の表出が有力である．語りに傾聴を示す目的で応答する発話（以下，傾聴応答）は，単に表出すればよいというわけではなく，適切なタイミングで表出する必要がある．本論文では，傾聴応答の自動生成の実現に向けて，傾聴応答の表出タイミングとしての適切さの度合いを推定する手法を提案する．本手法では，transformerベースの手法でエンコードされた語りの音響情報と言語情報を用いて，この度合いを推定する．本手法の推定性能を評価するために，10人の聴き手による傾聴応答131,616個からなる応答データを用いて実験を行い，本手法の有効性を確認した．また，この適切さの度合いを利用して表出タイミングを検出する実験を行い，その有用性を確認した．

DOI： 10.14923/transfunj.2022hap0013

Identifying Influential References in Scholarly Papers Using Citation Contexts

Bottom-up Japanese Word Ordering using BERT 査読有り

Masato Yamazoe, Tomohiro Ohno and Shigeki Matsubara

Proceedings of the 15th International Conference on Agents and Artificial Intelligence (ICAART-2023) 3 巻頁： 673 - 681 2023年2月

Japanese Word Reordering based on Topological Sort 査読有り

Peng Sun, Tomohiro Ohno and Shigeki Matsubara

Proceedings of the 15th International Conference on Agents and Artificial Intelligence (ICAART-2023) 3 巻頁： 768 - 775 2023年2月

Question Difficulty Decision for Automated Interview in Foreign Language Test 査読有り

Minghao Lin, Koichiro Ito and Shigeki Matsubara

Proceedings of the 15th International Conference on Agents and Artificial Intelligence (ICAART-2023) 3 巻頁： 768 - 775 2023年2月

Ikoma, T; Matsubara, S

LEVERAGING GENERATIVE INTELLIGENCE IN DIGITAL LIBRARIES: TOWARDS HUMAN-MACHINE COLLABORATION, ICADL 2023, PT II 14458 巻頁： 152 - 161 2023年

Citation count is commonly used as a straightforward metric for measuring the impact of a paper. However, since all citations are treated equally, citation count does not accurately capture the true influence of a particular cited paper on the citing paper. To accurately measure the individual impact of cited papers, it is required to identify those that have a high influence on a citing paper. This paper proposes a method to identify the influential citations using the text of citation contexts, specifically the citing sentences. Citing sentences contain the descriptions of the cited papers and the relationship between the citing paper and each cited paper. The proposed method extracts the descriptions of cited papers from the citing sentences and utilizes them to identify influential references. Experimental results have shown the benefits of using the extracted description of each cited paper.

DOI： 10.1007/978-981-99-8088-8_13

Question Difficulty Decision for Automated Interview in Foreign Language Test Open Access

Lin M., Ito K., Matsubara S.

International Conference on Agents and Artificial Intelligence 3 巻頁： 760 - 767 2023年

記述言語：英語掲載種別：研究論文（学術雑誌）出版者・発行元：International Conference on Agents and Artificial Intelligence

As globalization continues to progress all over the world, demand is growing for objective and rapid assessment of language proficiency in foreign language learners. While automated assessment of listening, reading, and writing skills has been proposed, little research has been done to automate assessment of speaking skills. In this paper, we propose a method of deciding the difficulty of questions generated for interview tests of skills assessment in speaking a foreign language. To address question difficulty flexibly according to the abilities of test takers, our method considers the appropriateness of responses from test takers. We implemented this method using the large-scale pre-trained language model BERT (Bidirectional Encoder Representation from Transformers). Experiments were conducted using simulated test data from the Japanese Learner’s Conversation Database to confirm the effectiveness of our method in deciding difficulty.

DOI： 10.5220/0011774900003393

Japanese Word Reordering Based on Topological Sort Open Access

Sun P., Ohno T., Matsubara S.

International Conference on Agents and Artificial Intelligence 3 巻頁： 768 - 775 2023年

記述言語：英語掲載種別：研究論文（学術雑誌）出版者・発行元：International Conference on Agents and Artificial Intelligence

In Japanese, some sentences are grammatically well-formed, but not easy to read. This paper proposes a method for Japanese word reordering, which first adapts a Japanese BERT model to predict suitable pairwise orderings between two words in a sentence, and then converts the predicted results into a graph. The vertices in the graph represent the words in the sentence. The edges represent the pairwise orderings between two words. Finally, topological sort is applied to find the correct word ordering in a sentence by visiting each graph vertex. We conducted an evaluation experiment with uneasy-to-read Japanese sentences created from newspaper article sentences.

DOI： 10.5220/0011775000003393

Bottom-up Japanese Word Ordering Using BERT Open Access

Yamazoe M., Ohno T., Matsubara S.

International Conference on Agents and Artificial Intelligence 3 巻頁： 673 - 681 2023年

記述言語：英語掲載種別：研究論文（学術雑誌）出版者・発行元：International Conference on Agents and Artificial Intelligence

Although Japanese is widely regarded as a relatively free word order language, word order in Japanese is not entirely arbitrary and has some sort of preference. As a result, it is an important technique to produce a sentence that is not only grammatically correct but also easy to read. This paper proposes a method for the word ordering of a whole Japanese sentence when the dependency relations between words are given. Using BERT, the method identifies the easy-to-read word order for a syntactic tree based on bottom-up processing. We confirmed the effectiveness of our method through an experiment on word ordering using newspaper articles.

DOI： 10.5220/0011743600003393

Extracting Information about Research Resources from Scholarly Papers 査読有り

Ayahito Saji and Shigeki Matsubara

Proceedings of the 24th International Conference on Asia-Pacific Digital Libraries (ICADL-2022) 2022年12月

Classification of URL Citations in Scholarly Papers for Promoting Utilization of Research Artifacts 査読有り

Masaya Tsunokake and Shigeki Matsubara

Proceedings of AACL-IJCNLP-2022 The 1st Workshop on Information Extraction from Scientific Publications 2022年11月

A Model-Theoretic Formalization of Natural Language Inference Using Neural Network and Tableau Method 査読有り

Ayahito Saji, Yoshihide Kato, Shigeki Matsubara

Proceedings of the 36th Pacific Asia Conference on Language, Information and Computation (PACLIC 35) 2022年10月

Construction of Responsive Utterance Corpus for Attentive Listening Response Production 査読有り

Ito K., Murata M., Ohno T., Matsubara S.

2022 Language Resources and Evaluation Conference, LREC 2022 頁： 7244 - 7252 2022年

担当区分：最終著者記述言語：英語掲載種別：研究論文（国際会議プロシーディングス）出版者・発行元：2022 Language Resources and Evaluation Conference, LREC 2022

In Japan, the number of single-person households, particularly among the elderly, is increasing. Consequently, opportunities for people to narrate are being reduced. To address this issue, conversational agents, e.g., communication robots and smart speakers, are expected to play the role of the listener. To realize these agents, this paper describes the collection of conversational responses by listeners that demonstrate attentive listening attitudes toward narrative speakers, and a method to annotate existing narrative speech with responsive utterances is proposed. To summarize, 148,962 responsive utterances by 11 listeners were collected in a narrative corpus comprising 13,234 utterance units. The collected responsive utterances were analyzed in terms of response frequency, diversity, coverage, and naturalness. These results demonstrated that diverse and natural responsive utterances were collected by the proposed method in an efficient and comprehensive manner. To demonstrate the practical use of the collected responsive utterances, an experiment was conducted, in which response generation timings were detected in narratives.

Extracting Information about Research Resources from Scholarly Papers

Saji, A; Matsubara, S

FROM BORN-PHYSICAL TO BORN-VIRTUAL: AUGMENTING INTELLIGENCE IN DIGITAL LIBRARIES, ICADL 2022 13636 巻頁： 440 - 448 2022年

This paper presents a method to extract information about research resource metadata from scholarly papers to expand research resource repositories using an academic knowledge graph. Here, we considered the hypothesis that information extracted from scholarly papers is beneficial to expand research resource repositories from two perspectives: (1) the amount of information in existing metadata, and (2) the number of entries in existing repositories. We constructed an academic knowledge graph, where the nodes and directed edges in the graph correspond to the entities and their relations in scholarly papers, respectively. To verify our hypothesis, we constructed a knowledge graph using 15,721 papers published in international conferences. We then investigated the expandability of a language resource metadata repository using the constructed knowledge graph. The experimental results demonstrated that the constructed knowledge graph could be used to enrich the descriptions of metadata and increase the number of entries in the repository.

DOI： 10.1007/978-3-031-21756-2_35

Construction of Responsive Utterance Corpus for Attentive Listening Response Production

Ito, K; Murata, M; Ohno, T; Matsubara, S

LREC 2022: THIRTEEN INTERNATIONAL CONFERENCE ON LANGUAGE RESOURCES AND EVALUATION 頁： 7244 - 7252 2022年

Estimating the Generation Timing of Responsive Utterances by Active Listeners of Spoken Narratives 査読有り

Koichiro Ito, Masaki Murata, Tomohiro Ohno, and Shigeki Matsubara

Proceeding of 2021 IEEE Automatic Speech Recognition and Understanding Workshop (ASRU 2021) 頁： 495 - 502 2021年12月

Using Citation Contexts in Scholarly Papers for Research Data Search 査読有り

Masaya Tsunokake and Shigeki Matsubara

Proceedings of the 16th International Joint Symposium on Artificial Intelligence and Natural Language Processing (iSAI-NLP 2021) 2021年12月

Identification of Important Utterances in Narrative Speech Using Attentive Listening Responses 査読有り

Shunsuke Kosugi, Koichiro Ito, Masaki Murata, Tomohiro Ohno, and Shigeki Matsubara

Proceedings of the 16th International Joint Symposium on Artificial Intelligence and Natural Language Processing (iSAI-NLP 2021) 2021年12月

A New Representation for Span-based CCG Parsing 査読有り

Yoshihide Kato and Shigeki Matsubara

Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing (EMNLP-2021) 頁： 10579 - 10584 2021年11月

Natural Language Inference using Neural Network and Tableau Method 査読有り

Ayahito Saji, Daiki Takao, Yoshihide Kato and Shigeki Matsubara

Proceedings of the 35th Pacific Asia Conference on Language, Information and Computation (PACLIC 35) 頁： 406 - 414 2021年11月

同時通訳における順送り方略の適用に影響する起点言語の特徴査読有り

蔡仲煕, 笠浩一朗, 松原茂樹

通訳翻訳研究 20 巻頁： 25 - 40 2021年2月

Word reordering and comma insertion integrated with shift-reduce dependency parsing 査読有り

Miyachi K., Ohno T., Matsubara S.

ICAART 2021 - Proceedings of the 13th International Conference on Agents and Artificial Intelligence 2 巻頁： 1144 - 1150 2021年2月

記述言語：日本語掲載種別：研究論文（国際会議プロシーディングス）出版者・発行元：ICAART 2021 - Proceedings of the 13th International Conference on Agents and Artificial Intelligence

Japanese has widely recognized as relatively free word order language. However, since Japanese word order is not completely arbitrary and has some sort of preference, even native Japanese writers often write Japanese sentences that are grammatically well-formed but not easy to read. Furthermore, in Japanese sentences, a comma plays an important role in explicitly separating the constituents such as words and phrases. This paper proposes a method of word reordering and comma insertion for hard-to-read Japanese sentences so that they become easier to read, as basic technique. Our contribution is to show the feasibility of concurrently performing word reordering, comma insertion, and dependency parsing.

A New Representation for Span-based CCG Parsing

Kato, Y; Matsubara, S

2021 CONFERENCE ON EMPIRICAL METHODS IN NATURAL LANGUAGE PROCESSING (EMNLP 2021) 頁： 10579 - 10584 2021年

Classification of URLs Citing Research Artifacts in Scholarly Documents based on Distributed Representations 国際誌

Tsunokake M., Matsubara S.

CEUR Workshop Proceedings 3004 巻頁： 20 - 25 2021年

記述言語：英語掲載種別：研究論文（学術雑誌）出版者・発行元：CEUR Workshop Proceedings

This paper describes methods for classifying URLs referring to research artifacts in scholarly papers, and examines their classification performance. The methods discriminate whether a URL refers to a research artifact or not and classify the identified URL into “tool” or “data.” The methods use distributed representations obtained from citation contexts of the URL. Each component of a URL can be regarded as a word, and the meaning of the entire URL can be generated by synthesizing the distributed representation of each component using compositional functions. This paper evaluates several types of compositional functions from the viewpoint of classification performance. Experiments with using URLs in international conference papers showed the effectiveness of our proposed compositional functions.

ESTIMATING THE GENERATION TIMING OF RESPONSIVE UTTERANCES BY ACTIVE LISTENERS OF SPOKEN NARRATIVES

Ito, K; Murata, M; Ohno, T; Matsubara, S

2021 IEEE AUTOMATIC SPEECH RECOGNITION AND UNDERSTANDING WORKSHOP (ASRU) 頁： 495 - 502 2021年

記述言語：英語掲載種別：研究論文（学術雑誌）出版者・発行元：2021 IEEE Automatic Speech Recognition and Understanding Workshop, ASRU 2021 - Proceedings

Narrating is a fundamental human desire. In order to satisfy the desire, it is more effective to have multiple listeners listen to narratives than to have one listener does. Conversational agents, e.g., communication robots and smart speakers, are expected to listen to narratives, instead of human listeners. For such agents to be recognized as narrative listeners, it is necessary to generate responsive utterances at appropriate timings. This paper proposes a method to estimate the generation timings of responsive utterances by active listeners of spoken narratives. The proposed method estimates them by encoding acoustic and linguistic information extracted from narratives with a transformer-based approach. To evaluate the performance of the proposed method, an experiment was conducted using 131, 616 responsive utterances to narratives. The experimental results demonstrated that the proposed method was effective for the estimation and both the acoustic and linguistic information was useful.

DOI： 10.1109/ASRU51503.2021.9688132

Natural Language Inference using Neural Network and Tableau Method

Saji A., Takao D., Kato Y., Matsubara S.

Proceedings of the 35th Pacific Asia Conference on Language, Information and Computation, PACLIC 2021 頁： 303 - 311 2021年

記述言語：英語掲載種別：研究論文（学術雑誌）出版者・発行元：Proceedings of the 35th Pacific Asia Conference on Language, Information and Computation, PACLIC 2021

Natural language inference (NLI) is the task of identifying the inferential relation between a text pair. Neural-based approaches recently achieved high performance on the NLI task, but they cannot explain their reasoning processes. On the other hand, symbolic approaches have an advantage that the reasoning process is understandable to humans. This paper proposes a method to integrate a neural NLI model and a tableau proof system, the latter of which explains the reasoning processes. The standard tableau method consists of decomposing logical formulas by applying inferential rules, and checking whether or not there exists a valuation that satisfies the given constraints. Unlike the standard tableau method, our method adopts not logical formulas but dependency structures as its ingredients and uses a neural NLI model for the latter process. In order to analyze the behavior of a neural NLI model, we conducted an experiment on the neural NLI model alone and the proposed method using the SNLI corpus.

A New Representation for Span-based CCG Parsing

Kato Y., Matsubara S.

EMNLP 2021 - 2021 Conference on Empirical Methods in Natural Language Processing, Proceedings 頁： 10579 - 10584 2021年

記述言語：日本語掲載種別：研究論文（学術雑誌）出版者・発行元：EMNLP 2021 - 2021 Conference on Empirical Methods in Natural Language Processing, Proceedings

This paper proposes a new representation for CCG derivations. CCG derivations are represented as trees whose nodes are labeled with categories strictly restricted by CCG rule schemata. This characteristic is not suitable for span-based parsing models because they predict node labels independently. In other words, span-based models may generate invalid CCG derivations that violate the rule schemata. Our proposed representation decomposes CCG derivations into several independent pieces and prevents the span-based parsing models from violating the schemata. Our experimental result shows that an off-the-shelf span-based parser with our representation is comparable with previous CCG parsers.

Word Reordering and Comma Insertion Integrated with Shift-Reduce Dependency Parsing

Miyachi, K; Ohno, T; Matsubara, S

ICAART: PROCEEDINGS OF THE 13TH INTERNATIONAL CONFERENCE ON AGENTS AND ARTIFICIAL INTELLIGENCE - VOL 2 頁： 1144 - 1150 2021年

What Affects the Word Order of Target Language in Simultaneous Interpretation

Cai Z., Ryu K., Matsubara S.

2020 International Conference on Asian Language Processing, IALP 2020 頁： 135 - 140 2020年12月

記述言語：日本語掲載種別：研究論文（学術雑誌）出版者・発行元：2020 International Conference on Asian Language Processing, IALP 2020

A simultaneous interpretation has to be conducted to shorten delay and reduce working memory load. The interpreters adopt a strategy that generates a target language in a word order which is similar to the source language. To make it clear how the interpreters decide the word order in practice, we conduct a statistical study based on the comparison of the word order between translation and simultaneous interpretation. The analysis reveals that in 10.5% of all cases, interpreters maintain the word order of source language, instead of adopting inversion as the translation, and that syntactic factors, including dependencies of chunk pairs, length of post-modifier, affect the word order decision more than non-syntactic factors.

DOI： 10.1109/IALP51396.2020.9310461

論文で引用された研究データの同定と分類

角掛正弥, 松原茂樹

電気学会論文誌Ｃ（電子・情報・システム部門誌） 140 巻 ( 12 ) 頁： 1357 - 1364 2020年12月

記述言語：日本語掲載種別：研究論文（学術雑誌）出版者・発行元：一般社団法人電気学会

This paper proposes a method for identifying and classifying the research data cited in scholarly papers, aiming at automatic generation of metadata stored in data repository. This study focuses on URL citations in the scholarly papers. That is, the targets are to identify the URLs referring to the research data and to classify them into tool and data. The method is realized as a multi-class classification (tool/data/others). The method acquires the distributed representations of the URLs from the context around them, and uses them as the input feature. There exists an advantage in that the meanings of URLs can be given based on their surrounding words. This study adopts an approach of computing the meaning of the entire URL from those of the components of the URL. In order to evaluate the performance of the proposed method, experiments on URL classification were conducted. The scholarly papers included in the proceedings of the international conference were used as experimental data. Experimental results have shown the effectiveness of the proposed method for identifying and classifying URLs referring to research data.

DOI： 10.1541/ieejeiss.140.1357

Identification of Research Data References Based on Citation Contexts

Parsing Gapping Constructions Based on Grammatical and Semantic Roles 査読有り

Yoshihide Kato, Shigeki Matsubara

Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing (EMNLP-2020) 頁： 2747 - 2752 2020年11月

Ikoma T., Matsubara S.

Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) 12504 LNCS 巻頁： 149 - 156 2020年

記述言語：日本語掲載種別：研究論文（学術雑誌）出版者・発行元：Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

In this paper, a method for the automatic identification of research data references in publications is proposed for automatically generating research data repositories. The International Conference on Language Resources and Evaluation (LREC) requires authors to list research data references separately from other publication references. The goal of our research is to automate the discrimination process. We investigated the reference lists in LREC papers and the citation contexts to find characteristic features that are useful for identifying research data references. We confirmed that key phrases appeared in the citation contexts and the bibliographical elements in the reference lists. Our proposed method uses the presence or absence of key phrases to identify research data references. Experiments on LREC proceedings papers proved the effectiveness of using key phrases in the citation context.

DOI： 10.1007/978-3-030-64452-9_13

Relation between degree of empathy for narrative speech and type of responsive utterance in attentive listening

Ito K., Murata M., Ohno T., Matsubara S.

LREC 2020 - 12th International Conference on Language Resources and Evaluation, Conference Proceedings 頁： 696 - 701 2020年

記述言語：日本語掲載種別：研究論文（学術雑誌）出版者・発行元：LREC 2020 - 12th International Conference on Language Resources and Evaluation, Conference Proceedings

Nowadays, spoken dialogue agents such as communication robots and smart speakers listen to narratives of humans. In order for such an agent to be recognized as a listener of narratives and convey the attitude of attentive listening, it is necessary to generate responsive utterances. Moreover, responsive utterances can express empathy to narratives and showing an appropriate degree of empathy to narratives is significant for enhancing speaker's motivation. The degree of empathy shown by responsive utterances is thought to depend on their type. However, the relation between responsive utterances and degrees of the empathy has not been explored yet. This paper describes the classification of responsive utterances based on the degree of empathy in order to explain that relation. In this research, responsive utterances are classified into five levels based on the effect of utterances and literature on attentive listening. Quantitative evaluations using 37,995 responsive utterances showed the appropriateness of the proposed classification.

同時通訳における順送り方略の適用に影響する起点言語の特徴査読有り

蔡仲煕, 笠浩一朗, 松原茂樹

通訳翻訳研究 20 巻 ( 0 ) 頁： 25 - 40 2020年

記述言語：日本語掲載種別：研究論文（学術雑誌）出版者・発行元：日本通訳翻訳学会

Simultaneous interpretation has to be conducted to shorten delay and reduce working memory load. Interpreters adopt a strategy that involves the generation of a target language in the word order that is similar to that of the source language. To clarify how the interpreters decide the word order in practice, we conduct a statistical study on the word order of simultaneous interpretation, based on comparison with written translation. The word order comparisons indicate that the strategy of maintaining the original word order is more likely to be adopted in cases where the word order is not consistent in different translations. We analyze the features of source languages, including dependencies of chunk pairs, length of post modifier, and distance between two chunks, and confirm that such features affect the word-order decision in simultaneous interpretation.

DOI： 10.50837/its.2002

PTB graph parsing with tree approximation 査読有り国際誌

Kato Y., Matsubara S.

ACL 2019 - 57th Annual Meeting of the Association for Computational Linguistics, Proceedings of the Conference 頁： 5344 - 5349 2020年

担当区分：最終著者記述言語：英語掲載種別：研究論文（国際会議プロシーディングス）出版者・発行元：ACL 2019 - 57th Annual Meeting of the Association for Computational Linguistics, Proceedings of the Conference

The Penn Treebank (PTB) represents syntactic structures as graphs due to nonlocal dependencies. This paper proposes a method that approximates PTB graph-structured representations by trees. By our approximation method, we can reduce nonlocal dependency identification and constituency parsing into single tree-based parsing. An experimental result demonstrates that our approximation method with an off-the-shelf tree-based constituency parser significantly outperforms the previous methods in nonlocal dependency identification.

Relation between Degree of Empathy for Narrative Speech and Type of Responsive Utterance in Attentive Listening

What Affects the Word Order of Target Language in Simultaneous Interpretation

Cai Zhongxi, Ryu Koichiro, Matsubara Shigeki

2020 INTERNATIONAL CONFERENCE ON ASIAN LANGUAGE PROCESSING (IALP 2020) 頁： 135 - 140 2020年

Ito, K; Murata, M; Ohno, T; Matsubara, S

PROCEEDINGS OF THE 12TH INTERNATIONAL CONFERENCE ON LANGUAGE RESOURCES AND EVALUATION (LREC 2020) 頁： 696 - 701 2020年

Parsing Gapping Constructions Based on Grammatical and Semantic Roles

Kato, Y; Matsubara, S

PROCEEDINGS OF THE 2020 CONFERENCE ON EMPIRICAL METHODS IN NATURAL LANGUAGE PROCESSING (EMNLP) 頁： 2747 - 2752 2020年

Parsing gapping constructions based on grammatical and semantic roles

Kato Y., Matsubara S.

EMNLP 2020 - 2020 Conference on Empirical Methods in Natural Language Processing, Proceedings of the Conference 頁： 2747 - 2752 2020年

記述言語：英語掲載種別：研究論文（学術雑誌）出版者・発行元：EMNLP 2020 - 2020 Conference on Empirical Methods in Natural Language Processing, Proceedings of the Conference

A gapping construction consists of a coordinated structure where redundant elements are elided from all but one conjuncts. This paper proposes a method of parsing sentences with gapping to recover elided elements. The proposed method is based on constituent trees annotated with grammatical and semantic roles that are useful for identifying elided elements. Our method outperforms the previous method in terms of F-measure and recall.

Statistical analyses of missing translations in simultaneous interpretation using a large-scale bilingual speech corpus

PTB Graph Parsing with Tree Approximation 査読有り

Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics (ACL-2019) 頁： 5344-5349 2019年7月

同時通訳における作業記憶の負荷と訳出との関係査読有り

笠浩一朗, 松原茂樹

通訳翻訳研究 ( 18 ) 頁： 1-12 2019年2月

同時通訳における語の欠落に影響を及ぼす要因の分析査読有り

蔡仲煕, 笠浩一朗, 松原茂樹

通訳翻訳研究 ( 18 ) 頁： 1-14 2019年2月

Cai Z., Ryu K., Matsubara S.

LREC 2018 - 11th International Conference on Language Resources and Evaluation 頁： 4282 - 4288 2019年

記述言語：英語掲載種別：研究論文（学術雑誌）出版者・発行元：LREC 2018 - 11th International Conference on Language Resources and Evaluation

This paper describes statistical analyses of missing translations in simultaneous interpretations. Eighty-eight lectures from English-to-Japanese interpretation data from a large-scale bilingual speech corpus were used for the analyses. Word-level alignment was provided manually, and English words without corresponding Japanese words were considered missing translations. The English lectures contained 46,568 content words, 33.1% of which were missing in the translation. We analyzed the relationship between missing translations and various factors, including the speech rate of the source language, delay of interpretation, part-of-speech, and depth in the syntactic structure of the source language. The analyses revealed that the proportion of missing translations is high when the speech rate is high and delay is large. We also found that a high proportion of adverbs were missed in the translations, and that words at deeper positions in the syntactic structure were more likely to be missed.

PTB Graph Parsing with Tree Approximation

Kato, Y; Matsubara, S

57TH ANNUAL MEETING OF THE ASSOCIATION FOR COMPUTATIONAL LINGUISTICS (ACL 2019) 頁： 5344 - 5349 2019年

Model-Theoretic Incremental Interpretation Based on Discourse Representation Theory 査読有り

Yoshihide Kato, Shigeki Matsubara

Proceedings of the 32nd Pacific Asia Conference on Language, Information and Computation (PACLIC 32) 頁： 1 2018年12月

Statistical Analysis of Missing Translation in Simultaneous Interpretation Using A Large-scale Bilingual Speech Corpus 査読有り

Zhongxi Cai, Koichiro Ryu, Shigeki Matsubara

Proceedings of the Eleventh International Conference on Language Resources and Evaluation (LREC 2018) 頁： 4282-4288 2018年5月

語りの傾聴を話し手に示す応答発話の収集

村田匡輝, 大野誠寛, 松原茂樹

電気学会論文誌Ｃ（電子・情報・システム部門誌） 138 巻 ( 5 ) 頁： 637 - 638 2018年5月

記述言語：日本語掲載種別：研究論文（学術雑誌）出版者・発行元：一般社団法人電気学会

This paper describes collection of responsive utterances of showing attentive hearing for narrative speech. In this research, a method of annotating an existing narrative corpus with responsive utterances was adopted. Actually, 16,213 responsive utterances were collected for a narrative corpus consisting of 13,381 utterances. It was confirmed that the method can collect a variety of natural responsive utterances.

DOI： 10.1541/ieejeiss.138.637

同時通訳における語の欠落に影響を及ぼす要因の分析

蔡仲煕, 笠浩一朗, 松原茂樹

通訳翻訳研究 18 巻 ( 0 ) 頁： 133 - 146 2018年

記述言語：日本語掲載種別：研究論文（学術雑誌）出版者・発行元：日本通訳翻訳学会

This paper describes corpus-based analyses of missing translations in simultaneous interpretations. We used English-to-Japanese interpretation data in the simultaneous interpretation database of Nagoya University for the analyses. The English lectures contained 46,568 content words, and 33.1% of them were missing in the translation. We analyzed the relations between omissions and the factors that influence them, i.e., the speech rate of the source language, delay of interpretation, part-of-speech, and depth in the syntactic structure of the source language. The analyses revealed that the proportion of missing translations is high when the speech rate is high, and the delay of interpretation is large, i.e., adverbs were missing in the translations in a high proportion; the words at deeper positions in the syntactic structure were more likely to be missed.

DOI： 10.50837/its.1808

同時通訳における作業記憶の負荷と訳出との関係国際誌

笠浩一朗, 松原茂樹

通訳翻訳研究 18 巻 ( 0 ) 頁： 147 - 158 2018年

記述言語：日本語掲載種別：研究論文（学術雑誌）出版者・発行元：日本通訳翻訳学会

By employing corpus-based analyses of English-Japanese simultaneous interpretation, we discuss the influence translations have on the working memory load of simultaneous interpreters. To conduct a quantitative analysis with a large corpus, we used the Simultaneous Interpretation Database of Nagoya University. We analyzed a total of 88 sets of simultaneous interpretations of 22 different lectures in English given by 17 interpreters. When the load on the working memory of simultaneous interpreters is increased, the speech time increases, the translation rate decreases, but the average speech rate does not exhibit a relevant change.

DOI： 10.50837/its.1809

Model-theoretic incremental interpretation based on discourse representation theory

Kato Y., Matsubara S.

Proceedings of the 32nd Pacific Asia Conference on Language, Information and Computation, PACLIC 2018 頁： 254 - 262 2018年

記述言語：英語掲載種別：研究論文（学術雑誌）出版者・発行元：Proceedings of the 32nd Pacific Asia Conference on Language, Information and Computation, PACLIC 2018

This paper proposes a model-theoretic approach to incremental interpretation where allsentence prefixes have semantic values. Theproposed semantics is based on DiscourseRepresentation Theory (DRT), where semantic representations (called DRSs) are interpreted as assignment updates. In our semantics, a partial DRS of a sentence prefix is interpreted as two sets which stipulate the assignment updates. One denotes possible updatesand the other denotes necessary updates. Withthe proposed semantics, we can assign truthvalues to sentence prefixes.

Statistical Analyses of Missing Translations in Simultaneous Interpretation Using a Large-scale Bilingual Speech Corpus

Cai, ZX; Ryu, K; Matsubara, S

PROCEEDINGS OF THE ELEVENTH INTERNATIONAL CONFERENCE ON LANGUAGE RESOURCES AND EVALUATION (LREC 2018) 頁： 4282 - 4288 2018年

Correcting Syntactic Annotation Errors Based on Tree Mining 査読有り

Kanta Suzuki, Yoshihide Kato, Shigeki Matsubara

IEICE Transactions on Information and Systems E100-D 巻 ( 5 ) 頁： 1106-1113 2017年5月

Collection of responsive utterances to show attentive hearing attitude to speakers

Ohno T., Murata M., Matsubara S.

Proceedings of the 11th International Conference on Ubiquitous Information Management and Communication, IMCOM 2017 ( 31 ) 2017年1月

記述言語：日本語掲載種別：研究論文（学術雑誌）出版者・発行元：Proceedings of the 11th International Conference on Ubiquitous Information Management and Communication, IMCOM 2017

One of the stressors for users who have a dialogue with a robot is a doubt about whether or not the robot can comprehend user's utterances. To solve this problem, it is effective for a robot to actively show the attitude of attentive hearing for user's utterances. This paper describes collection of responsive utterances to show the attitude of attentive hearing for the purpose of actualizing conversation robots with attitudes of attentive hearing. In the data collection, we made a worker respond to utterances in Japanese elders' narrative corpus so that the responses become appropriate as a role of a listener. As the result, we collected 4,885 responsive utterances uttered for narratives of 9 elders.

DOI： 10.1145/3022227.3022257

Left-Corner Parsing for Identifying PTB-Style Nonlocal Dependencies

対話コーパスを用いた相づち生成タイミングの検出査読有り

大野誠寛，神谷優貴，松原茂樹

電子情報通信学会論文誌 J100-A 巻 ( 1 ) 頁： 53-65 2017年1月

Kato Yoshihide, Matsubara Shigeki

自然言語処理 24 巻 ( 3 ) 頁： 371 - 394 2017年

記述言語：英語掲載種別：研究論文（学術雑誌）出版者・発行元：一般社団法人　言語処理学会

Nonlocal dependencies represent syntactic phenomenon such as wh-movement, A-movement in passives, topicalization, raising, control, and right node raising. Nonlocal dependencies play an important role in semantic interpretation. This paper proposes a left-corner parser that identifies nonlocal dependencies. Our parser integrates nonlocal dependency identification into a transition-based system. We adopt a left-corner strategy in order to use the syntactic relation c-command, which plays an important role in nonlocal dependency identification. To utilize the global features captured by nonlocal dependencies, our parser uses a structured perceptron. In experimental evaluations, our parser achieved a good balance between constituent parsing and nonlocal dependency identification.

DOI： 10.5715/jnlp.24.371