Automatic Construction of Multilingual Hyper Topic Map

National audienceInformation systems have become very complex because of the large volume of multidimensional and heterogeneous data they contain; the challenge is no longer to gather data but to extract and visualize relevant information. This paper explores the problem of modeling and knowledge representation using the Topic Map standard. Our work here focuses on how to automatically construct and update a navigable semantic structure based on the Topic Map model from multilingual and heterogeneous information sources. The approach presented in this paper has four phases: Resources structuring, Identification of topic and association types, Topic Map population and Topic Map visualization. Our approach makes advances in the following fields: we address heterogeneous, unorganized and multilingual information sources, we maintain an incremental aspect along all the Topic Maps building process, and provide a unifying portal for the presentation of the generated Topic Map to support multilingual environment and allow navigation through different users profile