Similarity in aligning Linked Open Data ontologies

Abstract

Traditional ontology alignment (OA) systems focus on aligning well defined and structured ontologies from the same or closely related domain to produce a set of equivalence mappings between concepts in the source and target ontologies. Linked Open Data (LOD) ontologies, however, present different characteristics from standard ontologies. For example, equivalence relations are limited among LOD concepts; thus for LOD ontology alignment, subclass and superclass mappings between the source and target should also be produced. Current research on aligning LOD ontologies relies on other ways to measure similarity between concepts than typically found in traditional OA systems. Enhancements made to an existing traditional OA system to enable LOD alignment and include the use of background knowledge such as Wikinet. Experiments using a set of LOD reference alignments to evaluate the enhanced OA system and their results are described. Their results demonstrate the enhancements improve the alignment these LOD ontologies.