A Markov random field approach for querying and reconciling heterogeneous databases

Abstract

The problem of defining a uniform query interface to access distributed heterogeneous data sources is becoming more and more important in the modern, distributed information systems. Main focus is on how to handle semantic heterogeneity of data in different sources to support integration and query processing activities. In both cases, techniques to assess similarity between schema elements of different data sources are required, to support the derivation of a reconciled representation at the global level and similarity-based search over different sources. In this paper we introduce an approach based on the Markov random field theory for similarity-based matching of database schemas.