[16p-K505-1~17] 23.1 Joint Session N "Informatics"

3:45 PM - 4:00 PM JST(6:45 AM - 7:00 AM UTC)

[16p-K505-11] Development of Automatic Suggestion System of Sample Information Data from Literature by Large Language Models

〇Tomoya Mato¹, Yu Takada¹, Masaya Kumagai^3,4, Yukari Katsura^1,2,3 (1.NIMS, 2.Univ. of Tsukuba, 3.RIKEN, 4.SAKURA Internet Inc.)

Keywords:

Materials Informatics (MI),Large Language Models (LLM),Database

Materials Informatics (MI) development requires large-scale datasets that reflect real-world materials research. To efficiently collect complex experimental data, including high-performance materials' fabrication methods and properties, we developed the Starrydata2 system, successfully gathering data from approximately 10,000 papers and 80,000 samples. Furthermore, we implemented automation for extracting sample information using external Large Language Models to enhance efficiency. This presentation will highlight our achievements and discuss prospects.