Text‐mined Assertion KP

Description: The Text-mined Assertion Knowledge Provider compiles assertions between pairs of biomedical entities mined from the available biomedical literature. Assertions are mined from sentences, and a sample of sentences that support each assertion are returned as metadata for each knowledge graph edge. Node types include chemicals/drugs, diseases/phenotypes, and genes/proteins. Predicates are modeled using the Biolink qualifier representation when possible, and represent the following relationships: treats, causes or contributes to, positively regulates, and negatively regulates. Note that text-mined assertions should be interpreted with some degree of skepticism as the underlying natural language processing algorithms used to extract the assertions can produce false positives.

Example edge: Metformin|CHEBI:6801 -- biolink:treats --> Type II Diabetes|MONDO:0005148

Data sources: Currently the Text-mined Assertion KP extracts assertions from titles and abstracts from PubMed and the PubMed Central Open Access subset of full text documents.

Regulatory requirement(s) and/or licensing restriction(s): Some of the full text documents from the PubMed Central Open Access Subset are licensed to non-commercial entities only.

Additional Links: Text Mining Provider home page