PDFDataExtractor

PDFDataExtractor extracts and interprets metadata from Portable Document Format (PDF) scientific literature to support chemical literature mining and metadata-driven analyses.

Key Features:

• Enhanced Data Extraction: Extracts latent metadata from PDF files and integrates chemical named-entity recognition via ChemDataExtractor.
• Template-Based Architecture: Uses a template-based architecture to reconstruct the logical structure of scientific articles and extract semantic information across diverse layouts.
• Comprehensive Metadata Extraction: Outputs detailed metadata in JSON and plain text formats, including paper title, authors, affiliations, email addresses, abstracts, keywords, journal names, publication years, DOIs, references, and issue numbers.
• Improved Precision: Demonstrates precision in extracting key metadata areas using a self-created evaluation article set.

Scientific Applications:

• Chemical literature mining: Enables extraction and correlation of chemical entities and property data from PDFs by leveraging ChemDataExtractor.
• Data-driven materials discovery: Provides structured metadata to support data science workflows and materials-discovery analyses driven by literature-derived data.

Methodology:

Uses a template-based architecture to reconstruct document logical structure, applies ChemDataExtractor for chemical named-entity recognition, outputs metadata in JSON and plain text, and assessed precision with a self-created evaluation article set.