A Novel Vertex Affinity for Community Detection
- Lawrence Livermore National Lab. (LLNL), Livermore, CA (United States)
We propose a novel vertex affinity measure in this paper. The new vertex affinity quantifies the proximity between two vertices in terms of their clustering strength and is ideal for such graph analytics applications as community detection. We also developed a framework that combines simple graph searches and resistance circuit formulas to compute the vertex affinity efficiently. We study the properties of the new affinity measure empirically in comparison to those of other popular vertex proximity metrics. Our results show that the existing metrics are ill-suited for community detection due to their lack of fundamental properties that are essential for correctly capturing inter- and intra-cluster vertex proximity.
- Research Organization:
- Lawrence Livermore National Lab. (LLNL), Livermore, CA (United States)
- Sponsoring Organization:
- USDOE
- DOE Contract Number:
- AC52-07NA27344
- OSTI ID:
- 1226940
- Report Number(s):
- LLNL-TR-677840
- Country of Publication:
- United States
- Language:
- English
Similar Records
A Novel Coarsening Method for Scalable and Efficient Mesh Generation
Application-Specific Graph Sampling for Frequent Subgraph Mining and Community Detection