Organisations are generally composed of several information systems which are key components in the decision making process. Finding specific functional information in these systems may prove to be difficult for non IT users. Data integration systems are used to centralise data from several information systems but the actual solutions simply expose the collected data and don't manage changes that may occur in the underlying data sources structure. The objective of our work is to express the collected data in terms of business knowledge and allow functional users to explore knowledge and retrieve the corresponding data without having to know the underlying data sources structures. These data are found in distributed and heterogeneous data sources belonging to different services or departments. Our proposal, KRISMAS, is based on ontologies modelling functional domains and a multiagent architecture performing the data retrieval and managing the changes that may occur within the data sources. By managing the changes we mean maintaining the ontologies coherent with the data sources.
