10 April 2018 A similarity based agglomerative clustering algorithm in networks
Author Affiliations +
Proceedings Volume 10615, Ninth International Conference on Graphic and Image Processing (ICGIP 2017); 106155I (2018) https://doi.org/10.1117/12.2302918
Event: Ninth International Conference on Graphic and Image Processing, 2017, Qingdao, China
Abstract
The detection of clusters is benefit for understanding the organizations and functions of networks. Clusters, or communities, are usually groups of nodes densely interconnected but sparsely linked with any other clusters. To identify communities, an efficient and effective community agglomerative algorithm based on node similarity is proposed. The proposed method initially calculates similarities between each pair of nodes, and form pre-partitions according to the principle that each node is in the same community as its most similar neighbor. After that, check each partition whether it satisfies community criterion. For the pre-partitions who do not satisfy, incorporate them with others that having the biggest attraction until there are no changes. To measure the attraction ability of a partition, we propose an attraction index that based on the linked node’s importance in networks. Therefore, our proposed method can better exploit the nodes’ properties and network’s structure. To test the performance of our algorithm, both synthetic and empirical networks ranging in different scales are tested. Simulation results show that the proposed algorithm can obtain superior clustering results compared with six other widely used community detection algorithms.
© (2018) COPYRIGHT Society of Photo-Optical Instrumentation Engineers (SPIE). Downloading of the abstract is permitted for personal use only.
Zhiyuan Liu, Zhiyuan Liu, Xiujuan Wang, Xiujuan Wang, Yinghong Ma, Yinghong Ma, } "A similarity based agglomerative clustering algorithm in networks", Proc. SPIE 10615, Ninth International Conference on Graphic and Image Processing (ICGIP 2017), 106155I (10 April 2018); doi: 10.1117/12.2302918; https://doi.org/10.1117/12.2302918
PROCEEDINGS
11 PAGES


SHARE
Back to Top