The Problem And Application Of Joint Recommendation In Event-based Social Networks

In recent years,the Internet has developed rapidly,the Internet technology has changed with each passing day,and social networks have emerged and become widespread.Event-based Social Networks(EBSN),a social network that is event-based,represented by Meetup,Plancast and Douban and combines online and offline models,has developed rapidly.Different from previous social scenes,users in event-based social networks can browse online group information to join in online interactions,and decide whether to participate in offline events based on online experience.However,novel social methods,huge user groups,massive event data,and complex behavioral interactions make it difficult for users to quickly find events and groups which attract users.Therefore,based on the above reasons,in order to give users a better experience,it is very urgent and necessary to study the recommendation strategy in EBSN.But,due to the special nature of EBSN,the traditional recommendation method for users only no longer applies.At the same time,in order to meet the various needs of users,multi-task recommendation should be implemented.The recommendation system should consider the characteristics of users,events and groups at the same time.In other words,while satisfying users' interests as far as possible,the global balanced distribution of features should be guaranteed and the interrelation between features should be fully calculated.Therefore,this paper has carried out research on the EBSN joint recommendation problem and application,which is summarized as follows:(1)This paper systematically introduces the recommendation system algorithm and text mining algorithm,introduces the new social network structure of the Event-based Social Network,and analyzes the significance of the current the event-based social network recommendation.This paper analyzes the research status and difficulties at home and abroad,discusses the advantages and disadvantages of different EBSN algorithms,and provides ideas and theoretical basis for the follow-up research.(2)For the complex behavioral interaction problem in Event-based Social Networks,this paper based on the factorization machine model,build models from user-event,user-group,content.Through LDA model extract the theme features of the text and the way to share the latent space is used,a Content-based Co-Factorization Machine(CC-FM)is proposed to improve the EBSN recommendation.(3)At present,the recommendation algorithms of Event-based Social Networks aremostly single user-event recommendations.However,Event-based Social Networks contain rich information and diverse functions,so that users are not only interested in events in the network,but also platform is required to provide user-group recommendation services.In order to better meet the various needs of users in social networks,the proposed algorithm implements a joint recommendation to implement user-event recommendation and user-group recommendation in an algorithm framework.Finally,this paper builds a test dataset on the real EBSN data according to the time-sensitive features of events in the Event-based Social Network,and carries out related contrast experiments of events and group recommendations to verify the model effect.Compared with the traditional recommendation algorithm,the proposed method has improved in many indicators.