Research On Link Prediction Method Combined Community Detection With Node Topological Structure

As an important research direction of social network analysis,link prediction is a method to predict potential links or missing links between nodes in the network based on the known network structure and the link relationship between the nodes in the network.With the widespread application of artificial intelligence and deep learning,link prediction has important research value in academic research such as predicting network evolution,knowledge graphs,and revealing protein interactions,as well as application services such as product recommendation and decision support services.At present,there are existing researches that incorporate community structure information into link prediction research,but the existing work lacks in-depth research on the integration of community structure information and link prediction.Therefore,this article mainly conducts link prediction work from the following two aspects on the basis of digging deep into the community structure information:1.Link prediction methods based on node similarity and link prediction methods based on local paths are two important types of link prediction methods.Existing link prediction methods based on node similarity usually consider the degree information of nodes in the social network;existing link prediction methods based on local paths mainly consider path information between nodes.Considering the advantages of the above two methods,this paper introduces the concept of local path on the basis of node degree,and proposes a link prediction method that combines community structure and local path similarity of node degree.Experimental results in common social networks and protein networks verify the effectiveness of this method.2.Existing link prediction methods based on community discovery usually only consider the influence of links between nodes in the same community,but do not distinguish the degree of influence on the links between nodes inside and outside the community.Therefore,this article first defines the intra-community correlation and inter-community correlation between nodes,and clarifies the impact of different community structures on the links between nodes.Furthermore,this paper proposes a link prediction method based on community discovery and node topology structure.Combining node topology information,improve the accuracy of link prediction methods.Experiments show that the method proposed in this paper can improve the effect of link prediction.The two methods proposed in this paper are based on community structure information,and improve the existing link prediction methods.Among them,the link prediction method that combines the local path similarity of the community structure and the node degree mainly considers the node degree and the local path relationship;the link prediction method based on community discovery and node topology structure mainly considers the relevance of links between nodes inside and outside the community.Both link prediction methods verify the effectiveness of the method in experiments.