Overlapping Community Detection And Robustness Analysis Of Complex Network Community Structure

In the real world,many systems can be modeled as complex network,such as social networks,transportation networks,computer networks,etc.Community structure is one of the most important properties of network.Structure analytics of complex networks can help to understand the functionalities and topology of real-world systems.Community detection has important meanings in the theory and broad application foreground,such as recommender systems,influence maximization.Network robustness is an important feature to measure the integrity of complex networks under attack.When the structural integrity of the network suffers attack,a part of system function will be lost.Studying the robustness of network can help us to build a stable real system for reduce malicious damage.In this thesis,we focus on the overlapping community structure of complex networks and the community robustness against bridge node attacks,the memetic algorithm and simulated annealing algorithm are applied respectively,The main work of this thesis consists of the following two parts:The overlapping community detection algorithm is used to find the overlap portion between communities in the network.In this thesis,based on modularity Q and general modularity density D,and the membership degree of every nodes in the network,we proposed the fuzzy evaluation index M,otherwise,this evaluation index can detect the different community structures of the network at different resolution.We use this evaluation index as an objective function to design a single objective optimization algorithm to realize community detection in the network.In the new algorithm,we redesign crossover and mutation operation,and use fuzzy k-means strategy as a local search strategy.Experiments on real networks show that our approach superior to modularity and general modularity density,and the algorithm are able to find overlapping communities in the network.The robustness of the network means the ability of keep structure integrity when the network fails or suffers attacked.Therefore,it is necessary to improve the robustness of the network.In this paper,we will model the robustness of community structure under most-visited bridge nodes attack,and propose a community optimization method based on simulated annealing.In this thesis,the traditional community detection algorithm BGLL was used to detect the community,and then a new way of the bridge node attack is proposed.The experiment shows that the largest connected component(LCC)and community integrity in the network more loss in the attack model.Then we proposed an index to evaluate community integrity under bridge node attack.Finally,a new constraint is added to the optimization algorithm,so that the optimized network is consistent with the community structure of the original network.Experiments show that the algorithm can effectively optimize the community robustness under the attack mode.