Research On Link Prediction Integrated With Community Structure Information And Node Information

With the development of Internet technology,link prediction,as one of the important methods for studying complex networks,has important theoretical and practical significance.In recent years,research results in this direction have emerged endlessly.However,the existing algorithms still have insufficient aspects in fully extracting network information and cannot accurately and efficiently predict missing links.In order to solve the above problems,this paper first uses the community detection algorithm to extract structural information in the complex network,and then integrates with the node information to make full use of the information in the complex network,and proposes a link prediction algorithm that integrates the community structure information and the node information.The contents are summarized as follows:1.Aiming at the problems of limited modularity resolution,low accuracy and slow convergence rate in traditional differential evolution,a community detection algorithm combining improved differential evolution and modularity density is proposed(Improved Differential Evolution and Modularity Density Community Detection,IMDECD).Firstly,the algorithm adjusts mutation strategy and parameters of differential evolution,and then the modularity density is used as the fitness function to overcome the limitation of the modularity resolution.Finally,the correction operation is performed according to the community structure to improve the individual quality in the population and accelerate the global convergence.2.To solve the problems that the existing link prediction algorithms take a long time or have inaccurate prediction results in large-scale networks,use the structural information in the community structure to integrate node similarity,and propose a link prediction algorithm that integrates community structural information and node information(Link Prediction Algorithm Integrated with Community Structure Information and Node Information,LPCN).Firstly,this paper defines a community similarity index based on the modularity density and the divided community structure;then this paper considers the node information in the network to calculate the node similarity;finally,the community structure information and the node information are integrated to obtain a fusion similarity index for link prediction,which can make full use of network information and improve the prediction accuracy of the algorithm.Finally,the feasibility of the proposed algorithm is verified through comparative experiments.The results show that the IMDECD algorithm has better accuracy and convergence performance,can obtain better community detection results,and can effectively extract structural information in the community structure.The LPCN algorithm that combines the extracted community structure information and node information for link prediction also has better prediction effect,taking into account the time complexity and prediction accuracy.