Semantic reconciliation in peer multi-data source management system

In Peer Data Management system (PDMS), two fundamental problems for data fusion arise: (a) how to build a semantic reconciliation between data sources schemas managed by peers, and (b) how to locate relevant peers for a given query. Our proposal lies in the application of multi-data source fusion approach [15] in the PDMS context. Multidata source schemas, which are distributed shared and maintained by peers, are the basis of a semantic overlay network. The design for Peer Multi-Data source Management System (PMDMS) was presented in [16] is an extension of MDS-Manager system [14][15] where data sources are distributed among peers. In this paper, we focus on the MatchMaker component in PMDMS that has the semantic reconciliation (i.e. mapping) responsibility between concrete data sources schemas (i.e. schemas describing data to share with other peers) also known as expertise. Our approach of semantic reconciliation is based on ontologies and XML technologies. Indeed, the peer schema (i.e. an ontology expressed with OWL/RDF), is annotated with a set of synonymous in order to guide later the search of semantics equivalences between expertises. The mappings results are stored in an XML document called Conflicts data source (a part of the multi-data source) as semantic links between concepts such as equivalence, synonym, homonym or disjoint concepts. Copyright l'2009 ACM.