Book Title Book Editors IOS Press, 2003 1 Using context to improve semantic interoperability Patrick Hoffmann , Lionel Médini and Parisa Ghodous 1 Lyon Research Center for Images and Intelligent Information Systems Claude Bernard Lyon 1 University, Villeurbanne, France Abstract. This paper presents an approach to enhance interoperability between heteroge- neous ontologies. It consists in adapting the ranking of concepts to the final users and their work context. The computations are based on an upper domain ontology, a task hierarchy and a user profile. As prerequisites, OWL ontologies have to be given, and an articulation ontology has to be built. Keywords. context, contextual ranking, semantic resources, semantic similarity 1. Introduction In an increasing number of organizations, virtual collaboration becomes a reality. More and more collaborative platforms provide means for connecting various models and tools used by different partners of Concurrent Engineering (CE) projects. Computerized data exchanges yet suffer from software incompatibilities, resulting in semantic losses when transmitting high level data. This is particularly true when considering complex data (3D data, simulation data, etc.). In CE, semantic resources(such as taxonomies, ontologies) are built for specific pur- poses, and evolve with the projects they are associated to. Semantic interoperability (i.e. interoperability between semantic resources) cannot therefore be achieved by integrat- ing ontologies but by establishing mappings between semantically related concepts from different ontologies. Since semantic interoperability depends on these mappings, it is essential to be capable of evaluating their relevance. As the same mappings are not as relevant for every users and every task, we herein propose an approach to compute a context-based evaluation of such mappings. Our approach is based on a context model composed of classifications of the organi- zation activity domains and tasks as well as of users’ profiles, on OWL ontologies, and on a ranking system. This ranking system receives requests on concepts from the OWL ontologies, made on users’ behalf. As results, it returns all semantically related concepts from the OWL ontologies, and ranks them according to the users’ contexts. This work is a continuation of Ferreira Da Silva et al.[1] contribution to semantic interoperability with SRILS, a middleware in which the ranking system is intended to be a module. 1 {patrick.hoffmann, lionel.medini, parisa.ghodous}@liris.cnrs.fr