Research And Implementation Of Context-aware Event Recommendation Algorithm In Event-based Social Networks

EBSN(Event Based Social Netwok)is a new kind of social network,which combines online virtual social relationships with real-world activities.EBSN provides a convenient platform for event organization,at the same time,it greatly reduces the difficulty of organizing activities in real life and facilitates people with common interests communicating face to face.There are sheer volume of events published in EBSN per day,which undermines the users' ability to choose the events that meet their needs.This problem caused the researchers'attention.Recommendation algorithms appear as a natural solution for this problem,but differently from classic recommendation scenarios,due to sparse RSVPs in EBSN and events are short-lived,the event recommendation problem is inevitably cold-start in EBSN.In order to solve the event recommendation problem in EBSN,this paper exploits several contextual signals available from EBSNs and proposes a context-aware event recommendation algorithm.Context-aware event recommendation algorithm combines users' RSVPs and social signals derived from group relationships,content signals derived from events5 description,events' location signals and events' temporal signals,tag signals derived from users and groups for learning to rank events for target users' personalized top-n recommendation.Target users' preferences and candidate events' features are obtained through analyzing the features of the above signals.Then this paper matches target users' features and candidate events'features,and designs a scoring function of multi-features to solve the cold-start problem of event recommendation caused by users and events in EBSN.Experiments using a large crawl of Meetup.com and Plancast.com demonstrate the effectiveness of Context-aware event recommendation algorithm in contrast to other event recommendation algorithms.The experimental results show that the algorithm proposed in this paper can solve the cold-start problem well,at the same time,it performs better on effectiveness compared with other event recommendation algorithms.Finally,this paper tests the effect of five signals on the effectiveness of the proposed context-aware event recommendation algorithm respectively with Meetup website's dataset.