Item-Based Top-k Influential User Discovery in Social Networks

Discovering top-k influential users plays a central role in many social network applications. In this paper, we study a challenging problem of discovering item-based top-k influential users in social networks. Specifically, we present a dynamic selection approach (referred to as Item-based top-K influential user Discovering Approach, IDA for short), to identify the top-k influential users for a given item based on real-world diffusion traces and on-line relationships. In particular, IDA first softly divides users involved in a diffusion trace into different communities by topic, and ranks users´ influence degrees in these topic communities with activeness, follower-counts, and follower participation-rates (including forwards and comments). In doing so, the top-K influential users for a given item can be obtained w.r.t. different topic communities. Experimental results on real world data sets demonstrate the performance of our approach.