Research And Implementation Of Community Detection Algorithm Based On Complex Networks

Research on complex networks is a hot topic in recent years.In real life,biological gene networks,internet of things,human relations networks,the Internet,etc.form a variety of complex network structures.The emergence of community detection is to observe the structure of complex networks and research the characteristics of complex network structures.Community detection is a process of transforming a disorderly network structure into a reasonable and effective network structure.Its processing objects are the connections,topology,and individuals of the network(abstracts are the edges,nodes,etc,of the network diagram),which has their own attributes(such as edge weights,direction,node degree of overlap,etc.).Community detection uses specific algorithms and the nature of the network to handle the process of community segmentation.The community structure of complex networks has a close relationship with the functions of the network(such as robustness,transitivity,etc.).Therefore,it is of great significance to find out the correct community structure of the network and analyze the relevant properties.Based on the point-to-point ratio,modularity community classification standards are currently the most commonly used measures to measure the quality of the network community structure,however,there are some problems that cannot be overcome,and there are also the resolution limited problems,it requires the improvement of methods to improve the accuracy of community detection.This paper proposes a new quality assessment method of community detection based on the joint evaluation of mutual information and information entropy.This method avoids the resolution limited problem well and maps the lossy topology compression process of the information transfer to the community detecting process.It uses the information volume calculation method to replace the modularity calculation method,which greatly improves the accuracy of the algorithm partitioning result.On this basis,the non-overlapping network community division method is further optimized,and the overlapping node judgment mechanism is added,so that it can be divided into communities for overlapping networkstructures,and a higher accuracy of the division results is obtained.The research in this paper reveals the deeper essential features of community division from the dual perspectives of mutual information combined with information entropy,and it is modeled and implemented to discover and reveal the rules among them,and is given the comparison and analysis of experimental results with other classical algorithms(such as GN algorithm,FastGN algorithm,LFM algorithm and CPM algorithm).