Faculty Profiles - MATSUBARA, Shigeki

Papers - MATSUBARA, Shigeki

Division display 1 - 20 of about 257 ／ All the affair displays >>

Generating Research Data Metadata from Their Accompanying README Files Reviewed

Kotaro Sekido, Yu Watanabe, Koichiro Ito, Shigeki Matsubara

Proceedings of the 3rd International Workshop on Natural Scientific Language Processing page： 181 - 185 2026.5

JSTS-Neg: Japanese Semantic Textual Similarity Dataset for Evaluating Negation Understanding Ability Reviewed

Reiko Yuasa, Yoshihide Kato, Shigeki Matsubara

Proceedings of the 15th Language Resources and Evaluation (LREC 2026) page： 12031 - 12041 2026.5

Identifying Implicit Research Data References in Paper Citations Reviewed

Koshi Motegi, Koichiro Ito, Shigeki Matsubara

Proceedings of the 3rd International Workshop on Natural Scientific Language Processing page： 186 - 192 2026.5

否定理解能力を評価するための日本語言語推論データセットJNLI-Neg Reviewed Open Access

吉田朝飛，加藤芳秀，小川泰弘，松原茂樹

情報処理学会論文誌 Vol. 66 ( 12 ) page： 1853 - 1860 2025.12

Where Did the Research Data Originate? Acquiring Provenance of Research Data from Scholarly Papers Reviewed

Koshi Motegi, Koichiro Ito, Shigeki Matsubara

Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) 2025.12

Dataset Similarity Estimation Using LLM-based Metadata Embeddings Reviewed

Koichiro Ito, Shigeki Matsubara

Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) 2025.12

Is CCGbank Semantically Valid? Insights from Negation Scope Analysis Reviewed

Kentaro Kojima, Yoshihide Kato, Shigeki Matsubara

Proceedings of the 38th Pacific Asia Conference on Language, Information and Computation (PACLIC-2025) 2025.12

Perspective-Aware Dataset Similarity Estimation Using Metadata Embeddings Reviewed

Koichiro Ito and Shigeki Matsubara

Proceedings of 1st Workshop on Smarter Extraction of ScholArly MEtadata using Knowledge Graphs and Language Models (SESAME2025) collocated with ACM/IEEE Joint Conference on Digital Libraries (JCDL2025) 2025.12

A Commonsense Knowledge Graph for Representing Exceptions Concerning Extrinsic Factors Reviewed

Shenghao Huang, Yoshihide Kato, Shigeki Matsubara

Proceedings of the 2025 IEEE International Conference on Big Data (IEEE BigData-2025) page： 5677 - 5686 2025.12

Metadata Generation for Research Data from URL Citation Contexts in Scholarly Papers: Task Definition and Dataset Construction Reviewed

Yu Watanabe, Koichiro Ito and Shigeki Matsubara

Proceedings of the 3rd Workshop on Artificial Intelligence for Scientific Publications (WASP2025) associated with the International Joint Conference on Natural Language Processing & Asia-Pacific Chapter of the Association for Computational Linguistics (IJCNLP-AACL) 2025.12

Dependency-aware Word Prediction Integrated with Incremental Parsing Reviewed

Hiroki Unno, Tomohiro Ohno, Koichiro Ito and Shigeki Matsubara

Proceedings of the 38th Pacific Asia Conference on Language, Information and Computation (PACLIC-2025) 2025.12

Development of Generative AI Chatbot forEnhancing Research Data Management in Universities Reviewed Open Access

Wada Kazuhiro, Matsubara Shigeki

Journal for Academic Computing and Networking Vol. 29 ( 1 ) page： 275 - 281 2025.11

Development of Organizational Research Data Management in Japanese Universities—through the Activities of the AXIES RDM Special Interest Group Invited Reviewed Open Access

Aoki Takaaki, Matsubara Shigeki

Journal for Academic Computing and Networking Vol. 29 ( 1 ) page： 1 - 6 2025.11

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 Vol. E108.D ( 10 ) page： 1183 - 1193 2025.10

Language：English Publishing type：Research paper (scientific journal) Publisher：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

Developing Rules and Guidelines to Support Research Data Governance in Universities Open Access

MATSUBARA Shigeki

The Journal of Information Science and Technology Association Vol. 75 ( 10 ) page： 522 - 526 2025.10

Language：Japanese Publishing type：Research paper (scientific journal) Publisher：Information Science and Technology Association, Japan

Research data management refers to the appropriate handling of information obtained through research activities. In recent years, the challenge has shifted from management entrusted to individual researchers or research groups to promoting systematic management under a research data infrastructure provided by universities. Such infrastructure encompasses not only technical components, such as data storage and authentication, but also rules and guidelines governing data handling. Research data governance at universities refers to the establishment of this infrastructure and the mechanisms by which members utilize and comply with it. Its realization directly contributes to improving research safety and efficiency. This paper introduces specific initiatives—such as developing implementation frameworks, formulating data policies, and guidelines—focusing on the case of Nagoya University.

DOI： 10.18919/jkg.75.10_522

Open Access

CiNii Research

Negation Scope Conversion for Unifying Negation-Annotated Datasets Reviewed Open Access

Yoshida Asahi, Kato Yoshihide, Matsubara Shigeki

Journal of Natural Language Processing Vol. 32 ( 1 ) page： 300 - 329 2025.3

Authorship：Last author Language：English Publishing type：Research paper (scientific journal) Publisher：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

Research data storage URL： https://github.com/asahi-y/negation-scope-conversion

Open Access

CiNii Research

A Preliminary Study on Metadata Compatibility and Automatic Conversion among LOD Datasets

Cong Y., Takaku M., Minaminyama Y., Aoki T., Matsubara S.

Proceedings of the International Conference on Dublin Core and Metadata Applications 2025

Language：English Publishing type：Research paper (scientific journal) Publisher：Proceedings of the International Conference on Dublin Core and Metadata Applications

In recent years, open science has promoted the web-based publication and reuse of research data. Publishing such data typically involves selecting a domain-specific metadata schema and using repositories for distribution. In Linked Open Data (LOD) datasets, domain-specific metadata schemas like VoID and DCAT are commonly used. However, differences in metadata schemas across institutions hinder cross-domain search and reuse due to duplicated terms and semantic ambiguity. The lack of clear metadata guidelines and examples leads to inconsistent metadata creation, ultimately compromising interoperability. This study addresses these issues by proposing a method to enhance compatibility among metadata schemas. Specifically, it focuses on mapping LOD schemas from VoID and DCAT to DataCite metadata schema. Additionally, it explores automating conversion process by pseudocode. This approach aims to align semantics across domain-specific metadata schemas and improve the findability and interoperability of dataset metadata.

DOI： 10.23106/dcmi.952519392

Scopus

A Commonsense Knowledge Graph for Representing Exceptions Concerning Extrinsic Factors

Huang S., Kato Y., Matsubara S.

Proceedings of the IEEE International Conference on Big Data Bigdata ( 2025 ) page： 5679 - 5688 2025

Language：English Publishing type：Research paper (scientific journal) Publisher：Proceedings of the IEEE International Conference on Big Data Bigdata

One of the features of commonsense knowledge is the existence of exceptions. While existing commonsense knowledge graphs (CKGs) can capture such relationships to some extent, their triple-based structure fundamentally limits the representation of exceptions influenced by extrinsic factors. The semantics of current triples assume uniform validity, making it difficult to express how commonsense knowledge may not always hold when affected by such extrinsic factors. This paper introduces the extrinsic factor-augmented commonsense knowledge graph (xCKG), which extends the conventional triple structure to a quadruple format that explicitly integrates extrinsic factors into the representation. Within the semantics of proposed xCKG, exceptions are formally defined by relating pairs of negated and affirmative quadruples through SubsetOf relation among the values of extrinsic factors. Specifically, when the value of one extrinsic factor is defined as a subset of another, the corresponding negated quadruple is regarded as an exception to the affirmative one, thereby enabling xCKG to formally capture exceptional relations as explicit structures within the graph. To confirm the feasibility of the proposed xCKG, we introduce a pipeline for constructing xCKGs from existing CKGs, utilizing large language models to generate and verify related tuples. An experiment was conducted to evaluate the pipeline, and the results demonstrate that the proposed pipeline can construct xCKGs from existing CKGs.

DOI： 10.1109/BigData66926.2025.11401422

Scopus

The 30th Anniversary Initiatives of the Association for Natural Language Processing Open Access

Inoue Naoya, Ouchi Hiroki, Kawahara Daisuke, Sudoh Katsuhito, Nagata Ryo, Komachi Mamoru, Kurohashi Sadao, Matsubara Shigeki, Miyao Yusuke

Journal of Natural Language Processing Vol. 32 ( 1 ) page： 330 - 341 2025

Human Performance in Incremental Dependency Parsing: Dependency Structure Annotations and their Analyses Reviewed

Hiroki Unno, Tomohiro Ohno, Koichiro Ito, Shigeki Matsubara

Proceedings of the 37th Pacific Asia Conference on Language, Information and Computation (PACLIC-2024) 2024.12

　 PREV - NEXT 　

<MATSUBARA, Shigeki>

Papers - MATSUBARA, Shigeki