Discovery Algorithm Based On Weighted Network Community Edge Clustering

There are different kinds of complex networks in the real world, such as socialrelations, the world wide web, protein networks, and community structure discovery is oneof the hot issues in the study of complex networks in recent years, the nodes and edges ofthe community can provide an abstract model to reflect the relationship between variousnetwork entities in the real world thus, society has practical significance for highertheoretical research method.Study on the community structure of the birth of a number of outstanding algorithms,such as hierarchical clustering algorithm, label propagation algorithm based on networktopology, the algorithm based on the. Found in the early community algorithm designseldom consider node overlap problem, resulting in not completely community algorithmresults consistent with the actual situation, such as hierarchical clustering GN algorithmbased on split betweenness, FN greedy algorithm and modularity based on incremental. Theadvantages of hierarchical clustering algorithm is a clear structure, drawback is that eachnode can only belong to a community, lost the overlapping of the account. The strength ofrelationship between nodes based on the relationship with the community network linkbetween the improved hierarchical clustering algorithm FN class cohesion, the relationshipsbetween the nodes into the relationship between the edges, and in order to more close to thereal world of the authenticity of the introduction of relationship strength node, consider anetwork with the right link relations based on LFN algorithm.Considering the hierarchical clustering algorithm can not find the defects ofoverlapping community structure, through the relationship between nodes into relationbetween edges on edge node for the line graph of agglomerative hierarchical clustering,because even the edge of the original node may belong to a single community with anumber of side connected from which community nodes belong to multiple communitiesdiscover overlapping communities, and the relationship between strength of nodes bynetwork join edges relationship. The LFN application and the real data set and comparedwith the traditional FN algorithm, the experimental data proves the feasibility of theimproved algorithm, and finally analyses the improved algorithm deficiencies may exist, theresearch and improvement to provide ideas for the following.