基于边聚类系数的谱聚类社区划分方法研究.

赵 菲; 余本国; 冀庆斌

As for the graph partition method， it can not get good partition effect when the network of community structure is not obvious， a spectral clustering community partition method based on edge clustering coefficient is proposed. Because the connection between nodes in the community is denser than that between nodes in each community， the size of edge clustering coefficient reflects the aggregation degree of nodes. Therefore， the clustering coefficient matrix is defined by the number of triangles constructed by edges in the community， and the elements in the matrix are the numbers of triangles actually formed by edges in the community. In the process of maximizing the aggregation degree function， the eigenvalues and eigenvectors of the matrix are used to divide the community. Experiments on real network data show that the algorithm is feasible.

COMMUNITY organization; EIGENVECTORS; TRIANGLES; COMMUNITIES; EDGES (Geometry)

Journal of Central China Normal University, 2020, Vol 54, Issue 1, p17

1000-1190

Academic Journal

10.19606/j.cnki.1000-1190.2020.01.004