| Abstract Scope |
Chemical reactions are central to materials synthesis and degradation. Machine reasoning enables expedited discovery in large parameter spaces with heterogeneous data. There are many challenges: lack of complete and open-source (meta)data, imbalanced data, lack of standards, and imprecise querying. We constructed a domain ontology (mds-ChemRxn) based on a comprehensive schema for reporting chemical reaction developed by Open Reaction Database (ORD) under MDS-Onto to describe the chemical reactions using Basic Formal Ontology (top-level ontology), Common Core Ontologies, and Chemical Entities with Biological Interest (mid-level ontologies) [1] . Historical data was transformed into Resource Descriptive Framework-star (RDF-star) (a graph data model) based on mds-ChemRxn. The resulting knowledge graph provides semantics and enables more accurate queries on chemical reactions by utilizing SPARQL and 1- and 2-hop reasoning. Searches for reactions with specific requirements (inputs, outputs, conditions, yields, etc.) become routine.
[1] [(B. Rajamohan et al., MDS-Onto, Scientific Data, doi: 10.1038/s41597-025-04938-5)] |