A Link Prediction Algorithm In Weighted Social Networks Based On Relationships Between Neighbor Nodes

As a kind of complex network,the application of social network is becoming more and more popular in our daily life.Link prediction in Social networks is one of the directions in social network analysis,it is becoming a hot research field in recent years.Link prediction aims to find links which actually exist,but for some reasons hidden away,or those which do not yet exist at present time,but will soon appear in the future.This research can help people understand the structure and evolution of the network and look for the potential relationships between organizations or individuals.Link prediction has been widely applied in electronic trading,friends recommending,biological networks and many other applications,so the study of link prediction in social networks makes a lot of sense.Existing link prediction algorithm mainly takes network's topology structure into account as a measure,but they usually only use some local characteristics of the network topology but fail to make full use of its characteristics,so the result of the prediction is not accurate.For example,the common neighbor algorithm simply refer to the number of common neighbors between two nodes,but ignore the relationship between the common node and non-neighboring nodes.To solve the above problem,in this paper we design a new algorithm named neighbor node relationship weight algorithm.This algorithm weights different edges between two nodes according to local characteristics of the network.The algorithm can distinguish contributions made by different paths to predict links.Then we take community structure into consideration to improve the result for link prediction,we can analyze that the location of common neighbor in different in the community has a certain impact on producing links between two nodes,and we weight different edges according to the community its two endpoints belong to.Finally,we regord a common neighbor community as a neighbor node and design a link prediction algorithm based on common neighbor community.Finally,the paper use social network datasets to verify neighbor node relationship weight algorithm.The experimental results show that new algorithm can not only keeps low time complcxity,but also improves the prediction accuracy.