Combining rough set theory and instance selection in ontology mapping

This paper presents a novel ontology mapping approach based on rough set theory and instance selection.In this approach the construction approach of a rough set-based inference instance base in which the instance selection (involving similarity distance, clustering set and redundancy degree) and discernibility matrix-based feature reduction are introduced respectively; and an ontology mapping approach based on multi-dimensional attribute value joint distribution is proposed.The core of this mapping approach is the overlapping of the inference instance space.Only valuable instances and important attributes can be selected into the ontology mapping based on the multi-dimensional attribute value joint distribution, so the sequenfly mapping efficiency is improved.The time complexity of the discernibility matrix-based method and the accuracy of the mapping approach are evaluated by an application example and a series of analyses and comparisons.