Tetsuji KUBOYAMA, Ph.D.
Professor,

Director, Computer Centre, &
Archival Science, Graduate School of Humanities,
Gakushuin University
1-5-1 Mejiro, Toshima-ku, Tokyo 171-8588, JAPAN [MAP]
E-mail:
Researchmap / ORCID / Google Scholar / DBLP

Current Research

Crystal growth, archival description and parliamentary text look unrelated, but they raise the same question: how do we enumerate discrete structures, and how do we compare them?

• Structure search on crystal surfaces. Enumeration of the discrete structures arising in crystal growth modelling, and similarity design over them. Recent work applies SAT solvers to validating the local electron counting rule, and combines density functional theory with Bayesian optimization to search for stable configurations.
• Fast search and clustering over large data. Filtering methods for nearest neighbour search using sketches, and clustering methods for large bipartite graphs.
• Text evaluation with large language models. Applying LLM-based evaluation to domains where judgements are obvious to humans but hard to produce consistently at scale, such as literary and political texts. One current target is the record of proceedings of the Japanese Diet.
• Computational archival science. Structural analysis of archival description standards — EAD, EAC-CPF and RiC — and extraction of context structures from descriptions.

See Research for details and CV for publications and appointments.

Open Teaching Materials

Teaching materials are published on a separate site.

• Records in Contexts: An Introduction

  A textbook on RiC, the ICA standard for archival description (PDF, CC BY 4.0). Also available in Japanese.

• Learning Data Analysis from Anti-Patterns

  Pitfalls of data analysis and machine learning, measured on a real authorship attribution task (PDF, CC BY 4.0, Japanese only). With marimo notebooks.

• Introduction to Computer Science

  Introductory lecture notes for undergraduates of any faculty, 13 chapters (HTML, CC BY-NC-SA 4.0, Japanese only).

News

• 2026.10 Sachiko Shimada, a graduate student in Archival Science, received the Excellent Presentation Encouragement Award at the 74th Annual Conference (2026) of the Japan Society of Library and Information Science (announcement, in Japanese), for her comparison of three ways of assigning subject terms to archival materials: NDLSH, free keywords and generative AI.
• 2026.10 Two graduate students in Archival Science, Atsuko Tsunoda and Sachiko Shimada, presented at the 74th Annual Conference of the Japan Society of Library and Information Science: on the potential users of archival institutions as seen in the Collaborative Reference Database, and on assigning subject terms to archival materials with NDLSH, free keywords and generative AI.
• 2026.10 A paper on approximate selection of the best matching unit for large-scale self-organizing maps, in the LNCS proceedings of SISAP 2026.
• 2026.10 A paper on memory-efficient approximate nearest-neighbour search via two-stage filtering, at KDIR 2026.
• 2026.09 A paper comparing the records disposition prescribed by government guidelines with what Japan’s central government agencies actually record, in Archival Science (open access).
• 2026.09 Records in Contexts: An Introduction updated to version 1.4.6 (Japanese) and 1.0.2 (English); revised editions will continue to appear.
• 2026.08 A paper on PyAPX, a Python toolkit for atomic configuration pattern exploration, in Scientific Reports (open access).
• 2026.08 The English edition of Records in Contexts: An Introduction is now available.
• 2026.06 Yuika Miura, a graduate student in Japanese Language and Literature, presented work on how generative AI affects the style of web novels at the 32nd Annual Meeting of the Association for Natural Language Processing (NLP2026) and the 40th Annual Conference of the Japanese Society for Artificial Intelligence (JSAI2026).
• 2025.10 Two papers on filtering methods for nearest neighbour search, in the LNCS volumes of ICPRAM 2024 and SISAP 2025.
• 2025.08 A paper on the search for stable atomic configurations in graphene-like BCN systems, in Crystal Growth & Design.
• 2025.01 A paper on a SAT solver-driven approach for validating the local electron counting rule, in the Journal of Crystal Growth.