一种基于规则学习的本体映射方法

Ontology mapping is an important way of establishing interoperabilitybetween (Semantic) Web applications that use differentbut relevant ontologies. In this paper, we propose a new approach forfinding complex semantic mappings across ontologies by transforming the problemto a rule learning task. Our approach, derived from the classical InductiveLogic Programming (ILP), generates mappings in the form of Hornrules. Specifically, it utilizes existing instance mappings as anchors betweenontologies to construct variable bindings for finding and validatingsemantic mappings. Different processing strategies are applied to classesand properties for improving the learning effectiveness. Our experimentalresults demonstrate the feasibility of our approach.