TSCDA: a dynamic two-stage community discovery approach

Gain and Pain in Graph Partitioning: Finding Accurate Communities in Complex Networks

“…(1) We synergistically integrate two highly effective community quality functions: (i) the connectivity-based metric introduced in [21] and (ii) the Max-Min Modularity [16]. Specifically, we employ the connectivity-based metric and the heuristic algorithm proposed in [21] to establish a novel complementary graph associated with the Max-Min Modularity and to obtain a high-quality initial solution to the Max-Min Modularity maximization problem. (2) We develop our community detection problem as an integer programming formulation of a revised Max-Min Modularity maximization problem (extending and improving the approach previously described in [15]).…”

“…In Section 2, we review the heuristic approach introduced in [21] and propose some theoretical insights regarding its efficiency. We then recap the traditional Modularity and Max-Min Modularity methodologies.…”

“…Heuristic techniques, on the other hand-like bottom-up (merging communities), top-down (splitting communities), and local search (migrating nodes)present viable solutions, especially for extensive networks [18], but they come at the cost of losing accuracy. Although various community discovery approaches have surfaced in recent years [19][20][21], due to space limitations, our discussion concentrates on methods pertinent to our study.…”

An Integer Programming Approach Reinforced by a Message-passing Procedure for Detecting Dense Attributed Subgraphs

“…More precisely, the local search algorithm works as follows: Randomly pick a subset of V as initial centers. As discussed above, move other vertices to these centers to form a solution (partitioning) and then compute the max-min modularity value Facebook348 [38] of the resulting partitioning; see (2). The local search technique tries to improve the max-min modularity value by adding and/or deleting a center to/from the set of centers at a time.…”

“…These subgraphs are well known as communities, and naturally, the process of identifying them is referred to as the community detection problem. Detecting communities has become one of the fundamental subjects in the field of network analysis and graph theory and has numerous applications in a wide range of areas, including the analysis of Social/Biological/Cosmological networks [1], [2], [3], [4] and WEB [5]. It also plays a crucial role in the domain of Network Design problems [6], Signal Processing [7], Image Segmentation [8], Pattern Recognition [9], and Data Mining [10].…”