ONTOLOGY MAPPING FOR INTEROPERABILITY IN SEMANTIC WEB Nuno Silva and João Rocha GECAD - Knowledge Engineering and Decision Support Research Group Instituto Superior de Engenharia do Porto 4200-072 Porto – Portugal Nuno.Silva@dei.isep.ipp.pt ABSTRACT Ontology mapping is the process whereby two ontologies are semantically related at conceptual level and the source ontology instances are transformed into target ontology entities according to those semantic relations. The objective of MAFRA–MApping FRAmework – is to cover all the phases of the ontology mapping process, including analysis, specification, representation, execution and evolution. The MAFRA Toolkit is an implementation of MAFRA, adopting an open architecture in order to observe the Semantic Web requirements, namely performance and transformation capabilities. One of the MAFRA Toolkit novelties respects its service-oriented approach, which claims that the capabilities of an ontology mapping system dependent on what transformations are present. Independent, plug able services are then responsible for the instances transformations, but they also provide support for other ontology mapping tasks like automatic specification of semantic relations, negotiation and evolution. KEYWORDS Semantic Web, Systems Integration, Ontology, Ontology Mapping 1. INTRODUCTION Ontologies, as means for conceptualizing and structuring knowledge, are seen as the key to the realization of the Semantic Web vision. Ontology allows the explicit specification of a domain of discourse, which permits to access to and reason about an agent knowledge. Ontologies raise the level of specification of knowledge, incorporating semantics into the data, and promote its exchange in an explicitly understandable form. Semantic Web and ontologies are therefore fully geared as a valuable framework for distinct applications, namely business applications like E-Commerce and B2B. However, ontologies do not overcome per se any interoperability problems, since it is hardly conceivable that a single ontology is applied in all kind of domains and applications. Ontology mapping does not intend to unify ontologies and their data, but to transform ontology instances according to the semantic relations (mapping relations) defined at conceptual level. Repositories are therefore kept separated, independent and distinct, maintaining their complete semantics and contents. The work described in this paper has been developed in scope of MAFRA-MApping FRAmework (Maedche et al 2002). MAFRA covers all the phases of the ontology mapping process, including analysis, specification, representation, execution and evolution. Current MAFRA implementation is MAFRA Toolkit. It adopts a declarative specification of mappings, hiding the procedural complexity of specification and execution, while its open architecture allows the integration of new semantic relations into the system, improving mapping capabilities as required. In Section 2, relevant background technology and projects in the area of knowledge interoperability will be analyzed and compared. Section 3 presents the MAFRA Toolkit service-oriented architecture. Section 4 describes the automatic semantic bridging process while section 5 describes and exemplifies the execution process. Section 6 describes the query web service, which allows independent agents to interoperate based on ontology mapping and Section 7 presents some experiences. Finally Section 8 provides an overview of the achieved results and point out some current and future efforts.