Interactive Refinement of Multi-scale Network Clusterings
Abstract
Insight of multiscale networks could be accessed through the visualization of automatic multiscale clusterings. But results of these methods do not necessarily fulfill user expectations since they don't provide error prone clusterings. In this article we propose a way to refine interactively these results by the use of multiscale grouping and ungrouping interactions. This approach revealed to give very good results on common networks, especially on small world networks. Moreover, the linear algorithm makes that the method remains interactive on huge graphs with thousand of nodes.