Dug

Dug provides semantic search across biomedical data repositories to improve findability and retrieval of relevant datasets by leveraging curated knowledge graphs and semantic web technologies.

Key Features:

• Semantic search using knowledge graphs: Dug uses evidence-based relationships from curated knowledge graphs to retrieve datasets relevant to queries including synonyms and related concepts.
• Semantic web technologies and knowledge graph methodologies: The system applies semantic web technologies and knowledge graph frameworks to link and contextualize dataset metadata.
• Enhanced recall versus Elasticsearch: In a comparative analysis Dug achieved total recall of 0.79 compared to Elasticsearch's 0.76 on a manually curated search dataset.
• Improved synonym and related-concept recall: For queries involving synonyms or related concepts Dug achieved total recall of 0.36 versus 0.14 for Elasticsearch, without sacrificing precision in top results.
• Extensive dataset indexing: Dug has indexed over 15,911 study variables from public datasets.
• Improves FAIRness: Dug enhances Findability, Accessibility, Interoperability, and Reusability (FAIR) of research data through semantic linking and discovery.

Scientific Applications:

• Cross-repository dataset discovery: Enables identification of relevant datasets across heterogeneous biomedical repositories using semantic relationships.
• Hypothesis formulation and testing: Supports formulation and testing of scientific hypotheses by broadening dataset retrieval through synonym and knowledge-graph-based linking.

Methodology:

Integration of semantic web technologies with knowledge graph frameworks to dynamically link datasets based on relationships defined by expert-curated ontologies and taxonomies.