Personalized Recommendation Methods Based On Community Overlap Coefficient In Social Networks

Due to the rapid development of social network in recent years,the research on social network becomes very meaningful.In this era of explosive growth of Internet information and data,users need to search and locate target data from massive information according to their customized needs.However,social networks,characterized by huge size and complex connections,are severe challenges users encounter in the process of social networking.In social networks,user information is inconvenient to obtain,and there is also a large amount of noise data.When user attribute information is missing,relevant models are created to excavate user's behavior data,so as to meet users' specific needs and realize personalized recommendation.This study highlights the importance and irreplaceability of recommendation system.At present,there are many mainstream recommendation algorithms researched at home and abroad,which belong to the category of strong relationship person recommendation algorithms.Recommendations based on similarity make users surrounded by homogeneous information,making it difficult to broaden the communication circle and obtain fresh information.According to Granonvetter's Weak relationship theory,weak relationship can often bring unique and interesting information,and weak relationship is more conducive to the diffusion of information.This article first briefly elaborates the definitions of strong and weak relationships in social networks,introduces the existing weak relationship person recommendation algorithm,and proposes the use of community overlap coefficients to identify strong and weak relationships for the identification of strong and weak relationships.According to the characteristics of the weak relationship,the weak ties are used to mine important nodes in the network,and the traditional recommendation algorithm is weighted to improve.Compared with the existing weak relationship recommendation algorithm,the recommendation effect of the algorithm proposed in this paper is significantly improved.According to the cascading effect in the social network,the choice of individuals will be influenced by the group,and two cascading recommendation algorithms are proposed from the two directions of community and neighbors.There are redundant edges in the network,which interferes with the calculation of the importance of the nodes in the network.An improved method is proposed for this problem,and the importance ranking is performed on the weak relation subnet.The improved importance ranking algorithm can more effectively mine nodes withstrong ability to get heterogeneous information,then recommend these nodes to the recommended user.Finally,this article uses the similarity index to restrict the recommendation algorithm,which is intended to give consideration to the possibility of making friends,and recommend more valuable links to the user.The experimental effect of the traditional recommendation algorithm is compared,in the collection of heterogeneous information,the recommend effect of the proposed algorithm is verified to be better.