Research On Event Recommender System Based On Social Network And Content

In recent years,with the progress and development of science and technology,the types and quantities of information have increased dramatically.How to get the information that users really need from the massive information has gradually become the focus of research.The recommendation system has gradually come into the attention of researchers.With the problem of information overload becoming more and more serious,event recommendation based on event-based social network has become a hot topic in current research.Different from recommendation in traditional fields,event recommendation based on social network environment has more problems.First of all,it is faced with the more severe problem of data sparsity,which makes the recommendation effect of traditional recommendation model average.Secondly,there exists heterogeneous social network information which cannot be used simultaneously in the general recommendation model.To solve the above problems and to improve the accuracy of recommendation results,this thesis proposes an event recommendation algorithm based on online social network and time information and a heterogeneous social network recommendation algorithm based on content information,and verifies its effectiveness on real data sets.Finally,this algorithm is applied to design and implement a community event recommendation system.The main research contents are as follows:(1)An event recommendation algorithm based on online social networks and time information is proposed.This recommendation algorithm uses matrix factorization algorithm as the scoring model to process the interaction information between users and events in event-based social networks.At the same time,it can be known from the analysis of existing data that there is a certain time rule for users to participate in offline events.In the real data,there is a strong time,so the time information is integrated into the preference score.According to the characteristics of real data of event recommendation,social relations and time characteristics are taken into account to improve the accuracy of recommendation.The experimental results on relevant real data sets show that compared with traditional algorithms,the proposed algorithm has improved the accuracy index and the normalized cumulative gain of loss(NDCG)index.(2)An event recommendation algorithm for heterogeneous social networks based on content information fusion is proposed.This algorithm uses Bayesian personalized sorting algorithm,which is more suitable for processing implicit information,as the scoring model.Considering the influence of event content information on recommendation performance in event-based social networks,feature extraction of event content information is carried out through long and short term memory network,which is commonly used in natural language processing field,and then integrated into the score.At the same time,in order to alleviate the problem of data sparsity,online and offline heterogeneous social networks containing characteristic information are integrated into the model through social regularization.Through the verification of Meetup City data set,the accuracy rate and other related indicators are improved.(3)A community event recommendation system based on social network is designed by using the event recommendation algorithm proposed in this thesis.According to the interactive information between users and events and the content information of events,the system can well recommend events that meet the interests of users.At the same time,the system basically realizes the basic functions required by users,and has passed the final test,to ensure the stability of the system.