Abstract:

Information retreival on the Web aims to provide to the user an easy access to information, that is located in a mass of textual documents. Our main contributions focus on improving the performance of information retrieval systems on the Web by increasing the selectivity of relevant documents. We propose a method of reformulation of the user query. The query will be pretreated and its named entities will be extracted. After that, the query is   semantically enriched using the Arabic Wordnet resource. This enrichment will improve the relevance between the documents of the corpus and the semantically enriched query. After extraction of the relevant documents from an expansion corpus these documents will be used for contextual enrichment of the initial query. In fact, we will return from these documents terms that have a contextual relations with those of the initial query through a Fuzzy association rule model. We will merge these two types of enrichment to obtain a well reformulated query