Abstract
In this document, we present research topic developed in the context of heterogeneous data integration in plant functional genomic. Plant functional genomic is a biological framework where several disciplines and techniques take part in the discover of genes function. It generates a large quantity of data which the scientists manage in various ways. However, many data sources, complementary or overlapping, are necessary to enrich information about genes function. The problem comes from the distribution, the autonomy and the heterogeneity of these sources. That drags biologists seeking information, to integrate results manually. The objective of this thesis is to make easier the scientists searches and to reach in a transparent way information resulting from several data sources. For that, we propose two approaches in order to evaluate the advantages and the disadvantages of them. Firstly we propose the integration of sources through the adaptation of a mediation system : Select. Successor of DISCO, Le Select allows the integration of heterogeneous and distributed data sources through a relationnal integration model. Secondly, we propose the creation of a user personalized environment that integrate data sources through workflows of Web services. This system is based on BioMOBY system and its Central Registry. To conclude, we propose a methodology adapted to the needs for similar integration projects.