GPT-Based Table Data Extractor
MaTableGPT is a GPT-based table data extractor for materials science literature, addressing previous limitations in handling diverse formats of reported tables. It achieves both high extraction accuracy (up to 96.8%) and low GPT usage cost (less than 6 US dollars) for the entire extraction processes over 10,000 articles. Finally, applied to literature of water-splitting catalysis, MaTableGPT enables large-scale data extraction and insightful statistical analyses. More details can be found in article number 2408221 by Anna Hiszpanski, Sang Soo Han, Donghun Kim, and co-workers.
Advanced Science, Volume 12, Issue 16, April 24, 2025. Read More
