Research On Community Evolution Based On CPM Algorithm

As one of the earliest research complex networks,DBLP scholar cooperation network provides a powerful support for related research.The research of scholars' cooperation network mainly includes the production of cooperation,the structural characteristics of cooperation formed,and the internal evolution mechanism of cooperation.By observing these characteristics,we can find some rules of scholars' cooperation network.The data volume of DBLP network branch is huge,and the overall research is difficult.The CPM is used to extend the core architecture of the algorithm to cut the network branches,so that all nodes have community ownership,and the number and size of the communities obtained are relatively reasonable.Calculate the inheritance relationship between communities and get the branch of the evolution of the community.In the course of the evolution of societies,there will be fusion and differentiation,in this paper,we mainly compare the stability and number of defferent structures with growth of the year,and scale growth of defferent structure societies in the process of evolution.First of all,according to the inheritance relationship of the community Between adjacent years,we can Get 9465 branchs of community evolution.Further more,by analyzing the branches of community evolution,we give the change of structure and scale in the evolution process of BA structure,and analyze the evolution characteristics of BA structure community.From the 3 BA communities of branch 12 in 1977 to the 3631 BA structure communities in 2010,there are 5320 non BA evolution into BA structural Societies,4494 times BA evolved into a non BA structure,19461 times to keep the structure of the BA community unchanged.After that,we select 80 evolutionary branches of the BA structure community,compare with initial 9 ER structure communities and 3 BA structure communities of DBLP evolution branches.The average growth rate and growth probability of the societies evolving into BA structure were 0.00297 and 0.491 respectively,which is lower than 0.297 and 0.574 of DBLP.the probability that the BA structure remains unchanged and the non BA structure evolves into BA structure is 0.925 and 0.443 respectively,higher than 0.812 and 0.142 of DBLP,societies with different starting structures will have an impact on the evolution structure and scale index of societies.Finally,we compare the average growth rate and growth probability of different structural communities,and the probability of structural evolution.The average growth rate of evolution to BA structure is 0.267,which is 6 times that of non BA structure,and the growth probability is 0.574,which is 24.1% higher than that of non BA structure 0.333.The probability of keeping the BA structure unchanged is 0.812 less than the highest regular structure,and the probability of a non BA structure evolving into a BA structure is 0.142,only 3.9% less than the highest ER random structure.The evolution data of different structural associations can conclude that the structure evolution probability of BA structure community is higher,and the average growth rate and growth probability of the BA community are higher than those of other structures.In the course of evolution,BA structure societies have the advantages of rapid structural growth,fastest scale growth and good structural stability.Through a large number of data set analysis o further understand the internal evolutionary relationship of DBLP scholars' cooperative networks.Predict the change of network structure.It is speculated that the evolution trend of the network will play a guiding role in scientific research activities,and even optimize the cooperation structure and promote the related development.