Automated materials synthesis requires historical data, but extracting detailed data and metadata from publications is challenging. In this project, we develop strategies for using large language models (LLMs) for rapid, autonomous data extraction from materials science articles in a format curatable by a materials database. We use the subdomain of polymer nanocomposites as our example use case and demonstrate a proof of concept case study via manual validation. We then develop a benchmark and a novel evaluation method to assess LLMs ability to extract structured data. We also use multi-modal approaches to extract valuable information.
Year
Student Trainees
Advisors