A Java framework for evaluating community mining algorithms. Which includes:
- Quality measures for a community structure (e.g. Q-modularity, Variance Ratio Criterion (VRC), Silhouette Width Criterion (SWC), Dunn index, etc.) presented in these paper:
- Agreement indices to compare two given community structures (e.g. Adjusted Rand Index, F-measure, Normalized Mutual Information, etc.), presented in following paper:
- R Rabbany et al., Generalization of Clustering Agreements and Distances for Overlapping Clusters and Network Communities; CORR 2014
- see supplementary materials
- see
measure\cluster - see a summary and examples in this ipython notebook
- Implementation of the TopLeaders algorithm presented in:
- R Rabbany et al., Top leaders community detection approach in information networks, SNA-KDD 2010
- see
algorithms\topleaders
- Different graph based distances and centrality measures
- see
measure\graph
- see
- Java wrappers for a collection of community mining algorithms
- see
algorithms\communityMiningandexecs\ - most of the executable files would need to be re-build on the specific machine
- see