Leveraging Hierarchy and Community Structure for Determining Influencers in Networks

Predicting influencers is an important task in social network analysis. Prerequisite for understanding the spreading dynamics in online social networks, it finds applications in product marketing, promotions of innovative ideas, constraining negative information etc. The proposed prediction method IPRI (Influence scoring using Position, Reachability and Interaction) leverages prevailing hierarchy, interaction patterns and community structure in the network for identifying influential actors. The proposal is based on the hypothesis that capacity to influence other social actors is an interplay of three facets of an actor viz. (i) position in social hierarchy (ii) reach to diverse homophilic groups in network, and (iii) intensity of interactions with neighbours. Preliminary comparative performance evaluation of IPRI method against classical and state-of-the-art methods finds it effective.