Clustering on dynamic social network data

Abstract

This paper presents a reference data set along with a labeling for graph clustering algorithms, especially for those handling dynamic graph data. We implemented a modification of Iterative Conductance Cutting and a spectral clustering. As base data set we used a filtered part of the Enron corpus. Different cluster measurements, as intra-cluster density, inter-cluster sparseness, and Q-Modularity were calculated on the results of the clustering to be able to compare results from other algorithms. © 2013 Springer-Verlag.