Complex concept matching for improved ontology alignment
Chaudhari, Tejas Dattatray
MetadataShow full item record
Modern ontology languages such as Web Ontology Language (OWL) allow defining complex concepts using restrictions, Boolean combinations, and exhaustive enumeration of instances. As ontology repositories proliferate on the Web, many contain ontologies that overlap in scope. Identifying this overlap is important for the discovery and exchange of knowledge that is related but modeled differently. Automated ontology alignment is becoming practical for finding such overlaps with minimal manual effort. Many of the current ontology alignment algorithms either do not consider the complex concepts in their alignment procedures or model them naively, thereby producing an alignment that is likely incomplete. We present an intuitive way to model complex concepts using their canonical subgraphs, and show how these may be integrated in multiple ontology alignment algorithms. The results indicate an overall improvement in the precision and F-measure of the alignment produced by these algorithms. However, this improvement is achieved at the expense of increased runtime due to the larger number of concepts that must now be considered by the alignment process.