Document analysis to provide semantic metadata based ontologies

Web services like conceptual search, i.e., search based on meaning rather than just character strings, has been the motivation of a large body of research in the IR field. The limits of information search engines called full-text depend mainly on the indexing process. Most approaches use free lexical tokens to annotate a given document, more, the selection of these entities is based on statistical and probability approaches such as the Vector Space Model (VSM) and the Latent Semantic Indexing (LSI). Although, some ones conclude acceptable results, the incidence of noise and silence affect severely the pertinence of returned results. The indexing process produce at the end set of independents terms (bag of words), without any semantic relationship. Our work proposes an approach to perform semantic annotation. We will use domain ontology to generate a set of structured meta data represented by a conceptual graph. It is a powerful formalism for representing rich ontological knowledge and allowing agents to reason about then through intelligent programs.