Research And Implementation Of Friend Recommendation Algorithm Based On Community Detection

In recent years,social network(Social Network Service,SNS)is a novel,practical and convenient way to make friends,it depends on the character of authenticity and stability,obtains users' favorite gradually.With the development of society and the progress of science and technology,a lot of iconic social networking sites have sprung up quickly,the information on the Internet has exploded and then people enter into an era with rich information and poor knowledge.Social networking sites with a huge amount of subscribers in order to better help users maintain the relationship between friends,get to know friends,expand their social range,the system must filter out a lot of redundant information.Under such background,the recommendation technology of friends came into being.Although many scholars have done a lot of research in the recommendation of friends,but the end result is not very perfect.Through observation and analysis in real social networking sites,we found that some users will gradually form a small group structure that is a community of a social network or you can understand it as the users 1 real social circle.In real life,most of us are active in these circles and the source of friends is mostly in this,so we put the community partition technology into the recommendation algorithm of the friends.The first part of the paper is the research on community discovery.The overlap community partition technology is still in the primary stage now,at present,the COPRA algorithm is a better algorithm.This thesis,I focus on the stochastic and the uncontainable scale problems of COPRA algorithm,we propose COPRA algorithm based on central point.The algorithm is divided into two steps,the first step is the selection of the center point based on User Influence Value,the second step is the process of performing COPRA algorithm using the central point as the seed node.In the design of the center point algorithm,calculating the influence of user nodes in social networks.According to the UIV values of the nodes that carry different tags,we choose the central nodes which represent the communities.So as to weaken the stochastic problem in the original algorithm,when the node selects its neighbor nodes we control the nodes to control the scale of communities.The second part is the research of the collaborative filtering recommendation algorithm,the introduction of community discovery technology reduces the size of the data.This process introduces the concept of association strength of user-label attribute to revise the user-label matrix.The process begins with the use of an improved approximate calculation method to determine the most similar Candidate items to the target item.We select the top K users as potential friends to recommend to the target user.Finally,this paper uses the Java language to implement the algorithm and tests these algorithms in the karate dataset and dolphin dataset and the user data sets from the website of Sina Weibo.Experimental results show that my algorithm is better than existing algorithms in the aspect of community discovery and recommendation of friends.