Arabic Information Retrieval System Based on Noun Phrases

In a rich information context, an information retrieval system must be able to ensure the best results. For this, the aim of our study consists in extracting the knowledge based on document textual contents by associating the analysis smoothness of a linguistic approach to the statistical approach capacity treating large corpus. The statistical approach is based on text mining, mainly on the latent semantic analysis technique; while the linguistic approach is based on the noun phrases which are more susceptible to be used like textual entities in representing the text information than the simple terms. By experimentation in Arabic documents, specialized in the environment field, we show the use of noun phrase impact on the information retrieval system precision.