Integrating Social Trust and Reputation in Resource Filtering and Recommendation Services
Résumé
The amount of data and resources available on the web to- day exceeds by far any other information repository in the history of human beings. The non-central nature of web re- sources allows these resources to keep expanding endlessly. This is manifested by the enormous amount of entries any simple query can return in any search engine, which makes it impossible to be utilized as a whole by an individual user. Many domains seek minimizing these returns by using dif- ferent methods, such as prioritizing relevant resources (in- formation retreival- recommender systems), or detecting ir- relevant resources (spam detection). Although these domains look far from each other in terms of techniques used and the way results are presented, we can safely say that they all share the same goal: Information filtering. In everyday life, reliability of information depends largely on the credibility of the sources from which they come. We tend to not question information that come from credible sources (people, TV Stations, Newspapers, websites...etc). Trust- ing a news outlet comes either from a personal experience (i.e.: previous interaction with this source), or from other sources citing this source as credible. This credibility can be a promising basis, on which we can build an effective filtering system. We take into account two levels of social trust: the relationship between information supplier and receiver, and the reputation of the supplier within the community. We are proposing an information filtering model that is aware of these two levels.