A Connexionist Model for Information Retrieval

This paper describes a connectionist architecture for an information retrieval system based on neural networks. This approach allows to define a “dynamic” thesaurus, in order to improve the construction of a documentary base and to perform associative information retrieval. We suggest a set of rules to activate cells in order to start an activation/propagation process on which the associative information retrieval is based. A learning mechanism is also started. These two notions allow to develop automatic reformulations of queries and dynamic restructuration of the information base.