The collaborative relevance in the distributed information retrieval

Abstract

Relevance is one of the most interesting topics in the information retrieval domain. In this paper, we introduce another method of relevance calculation. We propose to use the implicit opinion of users to calculate relevance. The Implicit judgment of users is injected to the documents by calculating different kinds of weighting. These latter touch several criteria like as user's weight in the query's words, user's profile, user's interest, document's content and the document popularity. In this method, each user is an active element of the system, he searches documents and he makes treatments to provide relevant information to other users in the Network. This is similar as the peer-to-peer systems; unlike that, an element (user) have to manage automatically his data by creating a short view model of his most visited documents, and calculates his relative relevance about each one. The relative relevance is variable according each user, so the final relevance is calculated by the averaging of the elementary relevance of all users. Hence, the name of collaborative relevance.