Analyzing tagged resources for social interests detection

Abstract

The social user is characterized by his social activity like sharing information, making relationships, etc. With the evolution of social content, the user needs more accurate information that reflects his interests. We focus on analyzing user's interests which are key elements for improving adaptation (recommendation, personalization, etc.). In this article, we are interested to overcome issues that influence the quality of adaptation in social networks, such as the accuracy of user's interests. The originality of our approach is the proposal of a new technique of user's interests detection by analyzing the accuracy of the tagging behaviour of the users in order to figure out the tags which really reflect the resources content. We focus on semi-structured data (resources), since they provide more comprehensible information. Our approach has been tested and evaluated in Delicious social database. A comparison between our approach and classical tag-based approach shows that our approach performs better.