From Mappings to Modules: Using Mappings to Identify Domain-Specific Modules in Large Ontologies

摘要

The problem of ontology modularization is an active area of research in the Semantic Web community. With the emergence and wider use of very large ontologies, in particular in fields such as biomedicine, more and more application developers need to extract meaningful modules of these ontologies to use in their applications. Researchers have also noted that many ontology-maintenance tasks would be simplified if we could extract modules from ontologies. These tasks include ontology matching: If we can separate ontologies into modules based on the topics that these modules cover, we can simplify and improve ontology matching. In this paper, we study a complementary problem: Can we use existing mappings between ontologies to facilitate modularization? We present a novel approach to modularization based on mappings between ontologies. We validate and analyze our approach by applying our methods to identify modules for National Cancer InstituteOs Thesaurus (NCI Thesaurus) and Systematized Nomenclature of Medicine-Clinical Terms (SNOMED-CT).