Comparing Louvain method and VOS Clustering

Comparison of results obtained by Louvain method and VOS Clustering are given by some examples. Only 10-100 restarts were used to obtain partitions, therefore it is possible that using larger number of restarts you get solutions with better Modularities or VOS Quality functions.
  1. Simple small example
  2. Constructed example with some explanations (provided by Ludo Waltman)
  3. Larger constructed example (provided by James Moody)
  4. Real example: Slovenian municipalities
  5. Another example

Final hint (given by Ludo Waltman): One possible advantage of the VOS clustering quality function over the modularity function is the elegant mathematical relationship between the VOS clustering quality function (used for partitioning the network) and the VOS mapping quality function (used for determining the layout of the network). The two quality functions can both be derived from the same underlying mathematical principle. Such a relationship is quite unique.

For details read the paper:
Ludo Waltman, Nees Jan van Eck, and Ed C.M. Noyons (2010): A unified approach to mapping and clustering of bibliometric networks. Journal of Informetrics, 4, 629-635

Back to Detecting Communities
Back to Pajek and Pajek-XXL Main page.