Ontological quality control in large-scale, applied ontology matching

Abstract

To date, large-scale applied ontology mapping has relied greatly on label matching and other relatively simple syntactic features. In search of more holistic and accurate alignment, we offer a suite of partially overlapping ontology mapping heuristics which allows us to hypothesise matches and test them against the knowledge in our source ontology (OpenCyc). We thereby automatically align our source ontology with 55K concepts from Wikipedia with 93% accuracy.