SERGE: Successive Event Recommendation Based on Graph Entropy for Event-Based Social Networks

Abstract

With the fast development of many event-based social networks (EBSNs), event recommendation, which is to recommend a list of upcoming events to a user according to his preference, has attracted a lot of attentions in both academia and industry. In this paper, we propose a successive event recommendation based on graph entropy (SERGE) to deal with the new event cold start problem by exploiting diverse relations as well as asynchronous feedbacks in EBSNs. The SERGE creates recommendation lists at discrete times during each publication period. At the beginning, it constructs a primary graph (PG) based on the entities and their relations in an EBSN and computes the user-event similarity scores by applying a random walk with restart (RWR) algorithm on PG. At each recommendation time, it then constructs a feedback graph (FG) based on the up-to-date user feedbacks on event reservations and applies the RWR again on FG to compute new user-event similarity scores. We then propose to weight the two sets of similarity scores with the graph entropies of both PG and FG and create the final recommendation lists accordingly. We have crawled two datasets from a real EBSN for two cities, Beijing and Shanghai in China. Experimental results validate the effectiveness and superiority of the proposed SERGE scheme over the peer schemes.