is a Data Source.
SemRep is a natural language processing program developed at the U.S. National Library of Medicine that extracts semantic predications (subject-relation-object triples) from biomedical text. It uses MetaMap to map free text onto Unified Medical Language System (UMLS) Metathesaurus concepts and the UMLS Semantic Network to assign relations, producing normalized assertions such as "Drug TREATS Disease". The semantic predications generated by SemRep over the PubMed corpus form the basis of SemMedDB, a PubMed-scale repository of biomedical predications widely reused for literature-based discovery and knowledge graph construction.
Unknown
Unknown
Unknown
| ID | Name | URL | Category | Format | Description |
|---|---|---|---|---|---|
| semrep.homepage | SemRep / SemMedDB SKR Tools Page | SemRep_SemMedDB_SKR.html | GraphicalInterface | http | SemRep and SemMedDB tool homepage and... |
| ID | Name | URL | Category | Format | Relation | Description |
|---|---|---|---|---|---|---|
| np-kg.graph.tsv | NP-KG TSV | NP-KG_v3.0.0.tsv?download=1 (1.0 GB) | GraphProduct | tsv | had primary source | Merged KG with ontology-grounded KG a... |
| np-kg.graph.networkx | NP-KG gpickle | NP-KG_v3.0.0.gpickle?download=1 (892.7 MB) | GraphProduct | mixed | had primary source | Merged KG with ontology-grounded KG a... |
SemRep is a rule-based semantic interpreter for biomedical text developed by the U.S. National Library of Medicine (NLM). It identifies UMLS Metathesaurus concepts in free text via MetaMap and then uses the UMLS Semantic Network, together with linguistic structure, to extract semantic predications: subject-relation-object triples such as “Indomethacin INHIBITS Cyclooxygenase”. Running SemRep across the PubMed corpus produces SemMedDB, a PubMed-scale repository of these predications that serves as an upstream source for many downstream knowledge graphs, including NP-KG.
Created: June 18, 2026 | Last modified: June 18, 2026