Community Detection Algorithm Based On Boundary Node And Label Propagation

At present,in order to enable people to find useful information more accurately and rapidly from the complex network,the researchers put forward the concept of community discovery.Community discovery can simplify the complex network structure,and facilitate researchers to analyze the complex network and explore the relationship among them.Therefore,how to find a reasonable and effective community structure is a popular research direction in recent years.Label propagation algorithm(LPA)is a classical algorithm to find the community,it is based on the idea of giving labels to the nodes in the graph,and then through the node label propagation iterative update,finally divides the nodes with a label to cover a community.The algorithm has been widely concerned by researchers because of its high time efficiency.However,there are many random steps in the process,and affect the stability of the final results.In order to improve the stability of label propagation algorithm effectively,this paper proposes a community discovery algorithm based on boundary node and label propagation,which is called LBN algorithm.LBN algorithm is mainly based on the following three key technologies:(1)The LBN algorithm in the initial stage to change the original LPA algorithm to label step for all nodes in the network diagram,but first of all to dig out the core node,then the core node gives the label,then step by step through a label transfer traverse to find the boundary nodes in the network diagram.This will reduce the number of iterative update labels in the original LPA algorithm,but also eliminates many of the random selection process,not only can improve the time efficiency of the iterative process,can also be divided into the initial number of communities.(2)Through the core node gives the weights to the boundary nodes,and selects the core nodes with large weights to cover the boundary nodes,when all nodes are labeled,Complete community partition.This step eliminates the label update process in the original algorithm to randomly select the label update steps,and to enhance the stability of the algorithm.In this paper,we select 2 data sets of the American political book network and the British university student association network,and test the results of LBNalgorithm and LPA algorithm on the two data sets.Finally,the tests of the two algorithms are compared.The results show that the LBN algorithm has a higher efficiency in maintaining the efficiency of the original LPA algorithm.