Integration of weighted knowledge bases

摘要

Integration of information from multiple sources is a key issue in many areas such as cooperative information systems, multi-databases and multi-agents reasoning systems, where information from different sources is often contradictory. In this paper we consider each information source to be a knowledge base with a weight representing the relative degree of importance of the source. We propose a formal semantics for merging multiple knowledge bases with weights. The semantics has desirable properties such as independence of the syntax forms of the knowledge bases and obeying the weighted majority rule in case of conflicts. We show by examples that this semantics returns intuitive results for the merging operation. We then present a syntactic characterization of the merging operation, which allows the result of merging to be obtained through a simple syntactic transformation of the knowledge bases.