Résumé
The FORUM project aims at extending existing data integration techniques in order to facilitate the development of mediation systems in large and dynamic environments. It is well known from the literature that a crucial point that hampers the development and wide adoption of mediation systems lies in the high entry and maintenance costs of such systems. To overcome these barriers, the FORUM project investigates three main research issues: (i) automatic discovery of semantic correspondences (ii) consistency maintenance of mappings, and (iii) tolerant rewriting of queries in the presence of approximate mappings.