A Normalized and a Hybrid Modularity
Haifeng Du,D. R. White,Y. Ren,Shuzhuo Li
2008-01-01
Abstract:The study of community structures is a hot spot for many inhomogeneous networks. Modularity plays an important role in this area, because it is a criterion for community detection, and a basis for community detection algorithms. Although commonly used in papers concerning community structures, modularity is seldom fully studied. In this paper, we investigate problems with the properties of modularity as defined by Newman and we propose a modularity normalized for number of groups as well as a hybrid modularity that improves on properties that reflect the interactions among communities. We also illustrate the basic flowchart of a “bottom-up merging” community detection strategy based on the properties of modularity, and explore a detection algorithm inspired by hybrid modularity. 1.Introduction The central idea of “Community Structure” is widely used in the study of social, biological and technical networks, among others. It represents an important future direction of complex networks research. In order to evaluate community structure, several measurements based on network density have been developed. [9] Modularity is the characteristic of a system that has been partitioned into smaller subsystems which interact with each other. The “modularity” index Q proposed by Newman et al. [10-14] is a popular method for evaluating how good is a particular division of networks into communities, assuming they interact but do not overlap. It is not only a criterion for community detection, but also provides basic insights needed to explore other algorithms for community detection. [1,10,14] There has been a recent proliferation of community structure studies and algorithms for fast computation of community structure in large networks. [14-15] Modularity usually acts as an objective function in these algorithms and thus community structure detection is basically an optimization process for the objective function Q. There are several other recent algorithms to detect community structure that are representative of the modularity approach . A network can be represented as G(V, A), where { } 3 2 1 , , v v v V = represents the set of nodes in the network and A is the set of connections. Detecting community structure consists of dividing a network into m sub-networks m p Vp , 2 , 1 , = ∅ ≠ ,