Mapping Composition for Matching Large Life Science Ontologies

There is an increasing need to interrelate different life science ontologies in order to facilitate data integration or semantic data analysis. Ontology matching aims at a largely automatic generation of mappings between ontologies mostly by calculating the linguistic and structural similarity of their concepts. In this paper we investigate an indirect computation of ontology mappings that composes and thus reuses previously determined ontology mappings that involve intermediate ontologies. The composition approach promises a fast computation of new mappings with reduced manual effort. Our evaluation for large anatomy ontologies shows that composing mappings via intermediate hub ontologies is not only highly efficient but can also achieve better match quality than with a direct matching of ontologies.