La pr\'ediction des int\'er\^ets des utilisateurs pour la RI contextuelle et la recommandation d'amis dans un environnement mobile

TL;DR

This paper presents a novel approach for contextual information retrieval and friend recommendation in mobile environments, leveraging user interests prediction from DBpedia and community discovery techniques.

Contribution

It introduces SA-IRI for predicting user interests based on context and Foaf-A-Walk for discovering communities for friend recommendations, combining semantic data and graph algorithms.

Findings

SA-IRI effectively predicts user interests in mobile contexts.

Foaf-A-Walk successfully discovers relevant communities for friend recommendations.

The approaches enhance personalization in mobile social applications.

Abstract

The emergence of smartphones has given mobile computing access to everyday reality. More specifically, the context modeling offers users an effective way to customize search results and even the recommended elements by limiting the data space. Moreover, in recent years, many social sites have embraced the notion of context in their recommendations. Indeed, with the availability of mobile devices, these new mobile sites have the advantage of providing users with more relevant elements based on their current situations. Thus, we introduce a new approach of contextual IR in a mobile environment. We offer a hand, an approach called SA-IRI based on the prediction of users' interests, from DBpedia, given their current situations. This approach applies the technique of associative classification in order to enrich the users' queries. Secondly, we introduce an approach of communities discovering, called Foaf-A-Walk, combining the random walk technique and the Foaf modeling, for friend recommendation.