Matching large ontologies: A divide-and-conquer approach

Ontologies proliferate with the progress of the Semantic Web. Ontology matching is an important way of establishing interoperability between (Semantic) Web applications that use different but related ontologies. Due to their sizes and monolithic nature, large ontologies regarding real world domains bring a new challenge to the state of the art ontology matching technology. In this paper, we propose a divide-and-conquer approach to matching large ontologies. We develop a structure-based partitioning algorithm, which partitions entities of each ontology into a set of small clusters and constructs blocks by assigning RDF Sentences to those clusters. Then, the blocks from different ontologies are matched based on precalculated anchors, and the block mappings holding high similarities are selected. Finally, two powerful matchers, V-Doc and Gmo, are employed to discover alignments in the block mappings. Comprehensive evaluation on both synthetic and real world data sets demonstrates that our approach both solves the scalability problem and achieves good precision and recall with significant reduction of execution time.