Research On Friend Information Service Recommendation Based On Community Division And User Similarity

With the rapid development of mobile Internet technology and the emergence of 5G networks,a variety of mobile social networking sites have emerged in large numbers and become an integral part of people’s lives and work.The role played by network users has also changed dramatically,and they have gradually changed from The recipient of the information becomes the publisher of the information.While mobile social networking sites bring countless conveniences and joys to everyone,it also leads to overload of information in mobile social networks.The massive and disorderly information in the network reduces the user experience and will inevitably lead to the continuous loss of users.Therefore,a recommendation system is generated.In the research of recommendation systems,friend recommendation and information service recommendation have become more important research directions in social networks.At the same time,the accuracy and efficiency of recommendation hinders the further development of the recommendation system.This article will study the problems existing in recommendation research.In mobile social networks,users will spontaneously form virtual communities of different sizes.Dividing online virtual communities can help with research on Internet public opinion monitoring,sentiment analysis,search engines,and recommendation systems.Community division is to divide user nodes in the network into different sets in order to ensure that the similarity of the internal nodes of the set is high and the similarity of the external nodes of the set is low.Most existing community partitioning algorithms are based on the similarity of user nodes,and lack of comprehensive research on the link relationship between nodes.In order to reasonably analyze the connection between users,the user relationship of mobile social networks can be described by introducing the degree of user trust to improve the quality of online community division.Secondly,the current recommendation algorithms are mostly based on user or item similarity for user recommendation or item recommendation.The similarity calculation dimension is relatively single and cannot adapt to large-scale network data recommendation research.On the other hand,recommendation algorithms focus on social relationships or hobbies,there are few recommendation algorithms that integrate social relationships and interests,and the accuracy of recommendation needs to be improved.This thesis focuses on the problems existing in the existing community partitioning algorithms and friend information service recommendation.Based on the characteristics of mobile social networks,this paper introduces user trust and integrates user similarity to propose a friend information service recommendation algorithm based on community division and user similarity.This recommendation algorithm introduces the degree of user trust to ensure the trustworthiness of community division,and then integrates user relationships and interest similarities to implement friend information service recommendation,which improves the accuracy of recommendation.The specific research contents of this article are as follows:First,the paper introduces the basic theory of mobile social networks,classic community division algorithm and friend information service recommendation algorithm.The representation form of the graph in the mobile social network is given;the relevant theoretical basis of the community and community structure are explained;and the advantages and disadvantages of the non-overlapping and overlapping community division algorithm are analyzed;the advantages and disadvantages of various friend recommendation algorithms are summarized,which lays a theoretical foundation for subsequent research in this paper.Secondly,this paper proposes a community partition method based on user trust and user similarity measurement method,and calculates user similarity based on the results of community partition to implement friend information service recommendation.In this article,the user’s trust level will be calculated by integrating the level of user interaction and the level of user expertise.In the measurement of user similarity,a method of integrating the two dimensions of user relationship and user interest similarity will be adopted.Finally,based on the results of community division,combined with user similarity help users find similar friends and the information they are interested in to achieve accurate and comprehensive service recommendation.Finally,based on the proposed friend information service recommendation algorithm of community partition and user similarity,we selected Weibo network data for empirical analysis.This paper proposes a new community partitioning algorithm based on the optimized modularity incremental algorithm,combining a fusion of user trust and the similarity of user nodes,and performs community partitioning on the obtained Weibo social network,and then analyzes and summarizes the partitioning results obtained based on the community partitioning algorithm in this paper.This paper uses three recommendation algorithm evaluation indicators: accuracy,recall,and F1 index to conduct a comparative analysis of this algorithm with traditional recommendation algorithms and mainstream recommendation algorithms,in order to verify the effectiveness of the community division on the recommendation algorithm and the reliability of this recommendation algorithm.