Multi-role event organization in social networks

Recently, event-based social networks (EBSNs) have become popular, hence how to organize a social event has received significant attention. Most of prior studies about social events organization usually consider the willingness of attendees and their relationships. However, they ignore the roles of attendees. In fact, many social events have requirement of attendees roles in the real world. In this paper, we propose to study the problem of Multi-Role Social Event Organization (MRSEO). Our goal is to maximize the overall harmony of the social event while considering multiple factors, such as attendees' roles, willingness and their relationships. To solve the problem, we propose two algorithms. Firstly, we propose a continuous relaxation technique based algorithm, called MRSEO-CRA. It converts the problem of MRSEO to an equivalent unconstrained continuous problem, and then employs RatioDCA algorithm to solve the converted one. Secondly, to better trade off between performance and running time, we further propose the other algorithm based on improved PageRank, called MRSEO-IPR. We conduct extensive experiments on real-world datasets to evaluate these two proposed algorithms and experimental results show that our algorithms outperform the state-of-the-art algorithm in terms of performance and running time.