“A Headline Generation Assistance Method for Manuals Using Large Language Models”

YAMAGATA Corporation will present its joint research with the Graduate School of Information Science and Technology at Hokkaido University (hereafter, Hokkaido University), titled “A Headline Generation Assistance Method for Manuals Using Large Language Models,” at the 24th Forum on Information Technology (FIT2025). The conference will be held September 3–5, 2025 at Hokkaido University of Science (Teine-ku, Sapporo).

Presentation Overview

Conference: The 24th Forum on Information Technology (FIT2025)
Dates: September 3 (Wed) – September 5 (Fri), 2025
Venue: Hokkaido University of Science (7-15-4-1 Maeda, Teine-ku, Sapporo, Hokkaido) / Hybrid format
Field / Session: Field O “Information Systems” — Session “Information Systems and Life Intelligence”
Presentation Time: September 5 (Fri), 13:10–15:40
Room: 7x
Paper No.: O-028 (7x-4)
Registration No.: 1088
Title: A Headline Generation Assistance Method for Manuals Using Large Language Models
Authors: Ryosuke Kamimae; Soichiro Yokoyama; Tomohisa Yamashita; Hidenori Kawamura (Hokkaido University); Isao Nagata (YAMAGATA)

General Program (Field O): https://www.gakkai-web.net/fit/program_web/data/html/program/o.html

Research Highlights

Purpose: Automatically generate a table-of-contents and headline structure from unstructured draft manuscripts so users can quickly reach target information, thereby reducing editorial workload.
Approach: Combine large language models (LLMs) with prompt engineering that reflects expert classification criteria, leveraging contextual understanding to produce user-centric structures.
Evaluation: In tests on product manuals, the generated outlines demonstrated structural validity and ease of information retrieval comparable to those created by human experts.

About FIT2025

FIT (Forum on Information Technology) is a scholarly forum co-hosted by the Information Processing Society of Japan (IPSJ) and the Institute of Electronics, Information and Communication Engineers (IEICE). The 2025 event will be held in a hybrid format at Hokkaido University of Science.

Official website: https://www.ipsj.or.jp/event/fit/fit2025/index.html

Researcher Profiles

The collaboration involves members of the Harmony Systems Engineering Laboratory: Professor Hidenori Kawamura, Associate Professor Tomohisa Yamashita, and Assistant Professor Soichiro Yokoyama.

Professor Hidenori Kawamura

His research focuses on artificial intelligence, multi-agent systems, complex systems engineering, and tourism informatics. He has authored numerous papers on AI and multi-agent systems and is a member of the IPSJ, JSAI, ORSJ, and the Society for Tourism Informatics (STI), where he also serves as a board member. He writes a regular column for Nikkei MJ titled “Kawamura Hidenori’s NARUHODO AI.”

Associate Professor Tomohisa Yamashita

His research focuses on artificial intelligence, multi-agent systems, social system simulation, and human-mobility analysis. He is a member of the IPSJ, JSAI, and ORSJ.

Assistant Professor Soichiro Yokoyama

His research focuses on combinatorial optimization and scheduling problems. He is a member of the IPSJ and ORSJ.

Research Institution

Name: Harmonious Systems Engineering Laboratory, Division of Computer Science and Information Technology, Graduate School of Information Science and Technology, Hokkaido University
Address: Kita 14, Nishi 9, Kita-ku, Sapporo, Hokkaido 060-0814, Japan
Research Areas: Artificial intelligence, deep learning, neural networks, machine learning, genetic algorithms, game theory, decision-making, multi-agent systems, data mining, and combinatorial optimization.

URL: http://harmo-lab.jp/