CRAFT

CRAFT provides a richly annotated biomedical text corpus with syntactic, formatting, and ontology-based semantic annotations to support natural language processing in biomedical literature.

Key Features:

• Manually Annotated Biomedical Corpus: Contains 67 full-text biomedical journal articles from the PubMed Central Open Access Subset with detailed manual annotations.
• Syntactic and Formatting Markup: Includes syntactic structure and document formatting annotations for linguistic analysis.
• Ontology-Based Semantic Annotation: Provides semantic annotations using the Uberon ontology for anatomical entities and life-cycle stages across species.
• Large-Scale Anatomical Concept Annotation: Offers extensive anatomical markup across full-text biomedical literature.
• Benchmark Dataset for NLP Systems: Serves as a gold-standard corpus for evaluating biomedical concept recognition and annotation systems.

Scientific Applications:

• Biomedical Natural Language Processing: Supports development and evaluation of algorithms for syntactic parsing, semantic analysis, and concept recognition.
• Automated Biocuration: Enables training of systems for automated extraction and curation of biomedical knowledge from literature.
• Ontology-Based Text Mining: Facilitates semantic integration of biomedical text annotations using anatomical ontologies.

Methodology:

CRAFT was constructed by manually annotating full-text biomedical articles with syntactic, formatting, and semantic annotations, including ontology-based annotations using the Uberon anatomical ontology.