Query modification based on relevance backpropagation
Mohand Boughanem, Chantal Soulé-Dupuy · 1997
It is well-known that relevance feedback has been an important method of improving the effectiveness of information retrieval systems. This observation is all the more true since information retrieval systems must get access to large document collections distributed over the world via international networks such as Internet. Relevance feedback can be viewed as an aid to the information retrieval task. In this paper, we present a relevance feedback strategy based on the backpropagation of the relevance of some initially retrieved documents by the use of an algorithm developed in a neural approach. Experiments performed with three collections show the effectiveness of this strategy.