Mining communities in networks: A solution for consistency and its evaluation

• Main Authors: KWAK, Haewoon, CHOI, Yoonchan, EOM, Young-Ho, JEONG, Hawoong, MOON, Sue
• Format: text
• Language: English
• Published: Institutional Knowledge at Singapore Management University 2009
• Subjects: community; modularity; cnm; wakita; louvain; social networks; consistent community identification; as graph; Numerical Analysis and Scientific Computing
• Online Access: https://ink.library.smu.edu.sg/sis_research/6102; https://ink.library.smu.edu.sg/context/sis_research/article/7105/viewcontent/1644893.1644930.pdf
• Institution: Singapore Management University

Online social networks pose significant challenges to computer scientists, physicists, and sociologists alike, for their massive size, fast evolution, and uncharted potential for social computing. One particular problem that has interested us is community identification. Many algorithms based on various metrics have been proposed for communities in networks [18, 24], but a few algorithms scale to very large networks. Three recent community identification algorithms, namely CNM [16], Wakita [59], and Louvain [10], stand out for their scalability to a few millions of nodes. All of them use modularity as the metric of optimization. However, all three algorithms produce inconsistent communities every time the ordering of nodes to the algorithms changes.We propose two quantitative metrics to represent the level of consistency across multiple runs of an algorithm: pairwise membership probability and consistency. Based on these two metrics, we propose a solution that improves the consistency without compromising the modularity. We demonstrate that our solution to use pairwise membership probabilities as link weights generates consistent communities within six or fewer cycles for most networks. However, our iterative, pairwise membership reinforcing approach does not deliver convergence for Flickr, Orkut, and Cyworld networks as well for the rest of the networks. Our approach is empirically driven and is yet to be shown to produce consistent output analytically. We leave further investigation into the topological structure and its impact on the consistency as future work.In order to evaluate the quality of clustering, we have looked at 3 of the 48 communities identified in the AS graph. Surprisingly, all have either hierarchical, geographical, or topological interpretations to their groupings. Our preliminary evaluation of the quality of communities is promising. We plan to conduct more thorough evaluation of the communities and study network structures and their evolutions using our approach.