Previously, 'related modules' were generated based on discussions in d.o. forum -- if several modules were mentioned in the same discussion threads, we consider them to be somewhat related. (More detailed explanation of the algorithms can be found in my previous Planet Drupal blogs).

We now developed a new algorithm that generates "related modules" based on project usage -- if several modules were enabled in the same site, then we consider them to be related. We got the project usage information from the "project_usage" module (http://drupal.org/project/usage) The core algorithm was based on the statistical concept, "correlation" (http://en.wikipedia.org/wiki/Correlation).

Usually, a Drupal site will only enable a set of modules that are complementary to each other, rather than a set of modules that substitute each other. Therefore, the new algorithm will have a bias towards generating more complements than substitutes. We don't know if that works better than the previous algorithms.

If everything goes well, the new algorithm will be enabled soon on d.o. We'll continue to use GA to see if the new algorithm works better or not.