The objective of the project is to develop an infrastructure for data integration in large scale distributed systems (such as the WWW) where data is stored in XML. This data integration should permit searches to be made both on keywords and on the contents of data. The principle of data integration is that the user's view of data should be uniform and transparent to the distribution and the heterogeneity of the data held in various sources. The topic of data integration is the subject of much international research, and many prototypes exist. The challenge today is the study of techniques of data integration to handle the loosely defined and rapidly changing nature of data on systems such as the Internet. In particular, handling data integration where there is no central authority to maintain the system, but where instead data integration is performed by different data sources negotiating with each other on a peer to peer (P2P) basis.
The XPeer architecture for P2P data integration based on the concept of super-peer was proposed by the team lead by Zohra Bellahsène at University of Montpellier. The XPeer approach makes it possible to combine the advantages of a hybrid P2P system (with some centralisation to assist with searches over the information) with that of a pure P2P system (giving better distribution of work, reliability, etc). The XPeer approach will be linked to the AutoMed data integration framework, which will provide the basic infrastructure for the data exchange between peers.