Avec le développement des supports de stockage, la quantité de documents disponibles
With the development of internet and storage devices, the quantity of available doc- uments increases quickly. It is necessary to have an information retrieval system able to ap- prehend efficiently these enormous quantities of documents. In this paper we propose DIIR (Dynamic Indexing for Information Retrieval), an information retrieval model based on the vector space model where the representation of documents is made dynamically in order to ex- pand requests and to adapt the term weights of documents. An evaluation of the model on real data from the evaluation campaign AMARYLLISS confirms the relevance of our approach.