研究者詳細 - 松原　茂樹

論文 - 松原　茂樹

分割表示 235 件中 21 - 40 件目／全件表示 >>

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

<p>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．</p>

DOI： 10.5715/jnlp.30.632

CiNii Research

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

CiNii Research

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

Identifying Influential References in Scholarly Papers Using Citation Contexts

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

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)

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

Web of Science

Scopus

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

Open Access

Scopus

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

Open Access

Scopus

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

Open Access

Scopus

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.

Scopus

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

Web of Science

Scopus

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

　前のページ - 次のページ　

<松原 茂樹>

論文 - 松原 茂樹

<松原　茂樹>

論文 - 松原　茂樹