Attribute-driven Topical Influential Users Detection in Online Social Networks

Abstract

At present, online social influencers are guiding the recognition and behaviors of their connections by becoming a voice of moulding opinions. As a consequence, influential user detection has become unavoidable to explore the dynamic evolution of Online Social Networks (OSNs) for any new procedure either for viral marketing applications or administrating the propagation of producing information. Existing methods pay less concentration on the temporal factor of the users’ interests. Our intent is to detect influential users who express their interest towards a particular query on multiple topics at various time periods by spotlighting more on users’ latest activities. The suggested temporal TwitterRank based topical influential users detection in multi hop neighbors network (TIUDMNN) method is based on the addition of PageRank algorithm. We also estimate the outcome of indirect influence i.e. focusing both on users’ influence to their direct neighbors and considering neighbors who are multi hops (2 or 3 hops) away. We conduct experiments on real datasets to illustrate the potency and performance of the proposed approach.