A Method to Manage the Difference of Precision between Profiles and Items for Recommender System - Applied Upon a News Recommender System using SVM Approach

Abstract

Contractors, commercial and business decision-makers need economical information to drive their decisions. The production and distribution of a press review about French regional economic actors represents a prospecting tool on partners and competitors for the businessman. Our goal is to propose a customized review for each user, thus reducing the overload of useless information. Some systems for recommending news items already exist. The usefulness of external knowledge to improve the process has already been explained in information retrieval. The system's knowledge base includes the domain knowledge used during the recommendation process. Our recommender system architecture is standard, but during the indexing task, the representations of content of each article and interests of users' profiles created are based on this domain knowledge. Articles and Profiles are semantically defined in the Knowledge base via concepts, instances and relations. This paper deals with the similarity measure, a critical subtask in recommendation systems. The Vector Space Model is a well-known model used for relevance ranking. The problematic exposed here is the utilization of the standard VSM method with our indexing method.