Research On Community Detection Algorithms In Complex Networks

Many complex systems in real life can be represented by complex networks,such as social networks,biological networks,information networks,transportation networks and so on.Most complex networks can be viewed as composed of multiple communities,and nodes belonging to the same community are more likely to have similar properties or functions.Community detection is to use the information contained in the network topology to resolve its modular community structure from the complex network.Community detection of complex networks is of great significance in analyzing the structure and function of networks,discovering the relationship between elements in networks,and predicting the dynamic development of networks.This thesis analyzes the development status of complex network community detection and the existing community detection algorithms,then tries to use a new method to solve the problem of community detection.This thesis proposes a community detection method based on node2 vec,which uses node2 vec algorithm as a bridge to convert community detection problems into vector clustering problems.The steps of the method are as follows: firstly,a second-order random walk strategy is adopted to generate a series of linear sequences;secondly,Skip-Gram model is used to train feature vectors;finally,clustering algorithm is used to cluster the trained node feature vectors to realize the division of communities.The feasibility of this method is proved by experimental comparison.For community detection of complex networks,algorithm based on modularity optimization has been a research hotspot this year.Adaptive genetic algorithm is a typical representative based on modularity optimization and has mature applications in community detection.However,the algorithm finally is easy to fall into local convergence,and the found good individuals are not the necessarily globally optimal one.In this thesis,an improved adaptive genetic algorithm is proposed.This algorithm designs a new adjustment formula for crossover mutation probability.The adjustment formula realizes the adaptive adjustment of the crossover mutation probability in all the domains of fitness,and segmentation.The floating range of the individual cross mutation probability is set,which not only improves the convergence speed of the population,but also retains the better individuals in the population.Through a series of experiments and comparative analysis with other algorithms,it is proved that the algorithm proposed in this thesis has greatly improved the accuracy of community detection.