Research And Application Of Community-based Critical Node Detection Model

In the complex network,the importance of each node is different and some critical nodes are more able to affect and control the spreading and structural characteristics of the entire network than other nodes.It's very meaningful for people to find these critical nodes so that limited resources can be invested in the appropriate area.The main research contents of this paper is described as follows:Local Fitness Method(LFM)may generate some homeless nodes because of its backtracking step.An improved LFM algorithm based on fitness function and community similarity was proposed to solve the problem.The new fitness value of nodes removes the backtracking step and considers the direct relationship,exclusive relationship,indirect relationship between the node and community.The direct relationship means the number of edges directly connected to the node and the community,exclusive relationship counts the communities which the node has already been adjacent to and indirect relationship is represented by the the remaining blank edges of the node.By locally optimizing the fitness function,the raw community partition is got,which may has some near duplicated communities.These duplicated communities need to be merged together by using the community similarity measure.The new community similarity criteria evaluates the similarity between two communities from the random graph based probability theory.It measures whether the relationship between the two communities is random by comparing the difference between real edges and expected edges between communities.The more it deviates from the random relationship,it means that the two communities are more similar.Gateway Local Rank(GLR)cannot distinguish the difference between the importance of nodes at the geometric center of the network and ignores the critical nodes located at the edge of the network,so a critical node detection algorithm based node interplay model(GLREX)was proposed to improve GLR.The node interplay model supposes the interaction force between nodes is proportional to the degree of nodes and inversely proportional to the exponential power of the distance.The interaction force between nodes is not necessarily inversely proportional to the square of the distance,which is different from the gravitational force between objects in the real world obeying the inverse-square law.Because the interaction force between nodes is transmitted through the shortest path between nodes,it means that the influence of nodes cannot be spread out uniformly in a circular manner to affect surrounding nodes.Therefore,the interaction force between nodes is not necessarily inversely proportional to the first or quadratic power of the distance.Because of the node interplay model,GLREX not only considers the interplay between nodes and communities but also the interplay between communities and communities,so that the critical nodes at different positions of the network can be effectively identified.By combining the improved community detection algorithm and the critical node identification algorithm,a community-based critical mode detection model was proposed.Then,the nuaa campus network was used as an example to verify the validity of the model.In addition,this paper also implemented an application software based on the model,including the overall design,module design,implementation of key technologies,etc.,and verified the feasibility of the model.