Presentation Information
[8p-P11-11]Materials Development with LLM (III): Extraction of Materials Data from Scientific Articles
〇Hiroyuki Oka1, Masashi Ishii1 (1.NIMS)
Keywords:
large language model,scientific articles,materials data extraction
Automatic extraction of materials data from scientific articles was examined using a Large Language Model (LLM). Gemma 3 was used as the LLM. Since Gemma 3 can also process images, the data extraction was performed from the main text and scatter plots of articles. The extraction was performed in the following three steps: (1) extraction of all sample names from main text, (2) extraction of preparation conditions and physical properties for each sample, and (3) data extraction from scatter plots. This presentation will report the details.