Modularity Function of Trapped-Probability for Network Clustering
The detection of community structure in networks is important because it gives insights into the structure–function relationship. The standard approach is based on the optimization of a relative quality measure, modularity, for the partition of a network. Modularity looks for regions of the network that have higher than expected number edges within them. We argue that the probability of a random walker being trapped in the original community can also give a measure of network connectivity. Based on this probability, we construct a generalized form of modularity to partition a network into communities by looking for regions of the network in which a random walker is more likely to stay. We evaluate our approach on two networks and show that it can effectively detect modules.
community random walk trapped-probability
Kun Zhao Quan Pan Shao-Wu Zhang
School of AutomationNorthwestern Polytechnical UniversityXi’an, Shaanxi, China School of Automation Northwestern Polytechnical University Xi’an, Shaanxi, China
国际会议
哈尔滨
英文
461-465
2011-01-18（万方平台首次上网日期，不代表论文的发表时间）