Several scientific projects focused on the creation of Peer-to-Peer data management system. The main objective of these systems is to allow data sharing and integration among a large set of distributed, heterogeneous data sources. The emergence of large scale systems provides solutions and brings to surface new challenging unsolved problems, among which, we address the data integration problem. In order to address this problem, we propose a new data integration approach that allows the semantic integration of heterogeneous and distributed data sources in a Peer-to-Peer environment with high distribution and evolution support. In this paper, we provide an introduction to the approaches; problems and research issues encountered when dealing with data integration.We present our approach and describe the several methods for constructing a global index that is the core of our approach by using semantic similarities. We end our work by an application example.