CeON/CERMINE

Extracts metadata, full text, and parsed references from academic PDFs using machine learning models (CRF-based parsers), outputting results in NLM JATS format. Available as a standalone JAR, Maven library, or REST API, supporting batch processing of documents and granular extraction of specific components like reference strings and affiliation data. Built in Java with learned models for document structure analysis and information extraction from scholarly publications.

513 stars. No commits in the last 6 months.