A Label Propagation Algorithm Combining Eigenvector Centrality and Label Entropy

Shucan Pan, Wenchao Jiang, Sui Lin, Dongjun Ning

Abstract: Aiming at the strong randomness and low accuracy problem caused by fuzzy boundaries of overlapping communities, a label propagation algorithm combining eigenvector centrality and label entropy (ECLE-LPA) is proposed. The K-kernel iteration factor and the eigenvector centrality of the node are used to calculate the node influence. In the propagation process, the label entropy and the closeness of the node are calculated to update the node label list, and the corresponding label memberships in this label list. These can overcome the overlapping community fuzzy boundaries recognition problem. The experimental results show that in the real network such as Les, Pollbooks, Football, Polblogs, Netscience, the EQ value of ECLE-LPA algorithm is generally increased by 1%–3% compared with the contrast algorithm. In the artificial network with fuzzy community structures, the NMI value of ECLE-LPA is more than10% higher than the contrast algorithm.