Multi agent celebrity recommender system (MACeRS): Twitter use case.

Mir Saman Tajbakhsh, Farnaz Ayough, Vahid Solouk, Hanif Emamgholizadeh, Jamshid Bagherzadeh.

Abstract: The advancement of social communities and virtual interaction of thoughts has apparently made social networking one of the fastest growing concepts. The interaction carries meanings beyond friendship and is applied to larger areas such as communities and networks for business, trade, cinema, and broadcasting. In a social network the user desire to find her or his interests which this issue plays a major part in initiation and growth of the community. However, the lack of important and useful information, and sometimes its inaccessibility, hinder users of establishing good connections, and thereby, expanding the community. The current paper presents a method of friend recommendation system based on the preferences and tendencies of the user and his or her friends. The proposed method introduces a novel way of extracting and modeling recommendation process as a game theory problem with two main agents (Celebrity and Non-Celebrity) for selecting the members to be recommended. We have used the real data from Twitter social network celebrity members in order to test and analyze our proposed system from two aspects, i.e., recommender system and social network, and then, discuss the results.

Share this page: