Enhancing Community Detection By Affinity-based Edge Weighting Scheme
- Lawrence Livermore National Lab. (LLNL), Livermore, CA (United States)
Community detection refers to an important graph analytics problem of finding a set of densely-connected subgraphs in a graph and has gained a great deal of interest recently. The performance of current community detection algorithms is limited by an inherent constraint of unweighted graphs that offer very little information on their internal community structures. In this paper, we propose a new scheme to address this issue that weights the edges in a given graph based on recently proposed vertex affinity. The vertex affinity quantifies the proximity between two vertices in terms of their clustering strength, and therefore, it is ideal for graph analytics applications such as community detection. We also demonstrate that the affinity-based edge weighting scheme can improve the performance of community detection algorithms significantly.
- Research Organization:
- Lawrence Livermore National Lab. (LLNL), Livermore, CA (United States)
- Sponsoring Organization:
- USDOE
- DOE Contract Number:
- AC52-07NA27344
- OSTI ID:
- 1226950
- Report Number(s):
- LLNL-TR-677818
- Country of Publication:
- United States
- Language:
- English
Similar Records
Tolerating the community detection resolution limit with edge weighting.
Scalable Pattern Matching in Metadata Graphs via Constraint Checking