6533b7dcfe1ef96bd1273094

RESEARCH PRODUCT

Evaluation de la pertinence dans un système de recommandation sémantique de nouvelles économiques

subject

description

Today in the commercial and financial sectors, staying informed about economic news is crucial and involves targeting good articles to read, because the huge amount of information. To address this problem, we propose an innovative article recommendation system, based on the integration of a semantic description of articles and on a knowledge ontological model. We support our recommendation system on an intrinsically efficient vector model that we have perfected to overcome the confusion existing in models between the concepts of similarity and relevancy that does not take into account the effects of the difference in the accuracy of the semantic descriptions precision between profiles and articles, on the perceived relevancy to the user. We present in this paper a new evaluation of the relevancy adapted to vector model.