Evidential Independence Maximization on Twitter Network

Detecting independent users in online social networks is an interesting research issue. In fact, independent users cannot generally be influenced, they are independent in their choices and decisions. Independent users may attract other users and make them adopt their point of view. A user is qualified as independent when his/her point of view does not depend on others ideas. Thus, the behavior of such a user is independent from other behaviors. Detecting independent users is interesting because a part of them can be influencers. Independent users that are not influencers can be directly targeted as they cannot be influenced. In this paper, we present an evidential independence maximization approach for Twitter users. The proposed approach is based on three metrics reflecting users behaviors. We propose an useful approach for detecting influencers. Indeed, we consider the independence as a characteristic of influencers even if not all independent users are influencers. The proposed approach is experimented on real data crawled from Twitter.