I've been wondering how to

I’ve been wondering how to automate the discovery of groups that people clump into while online. Here in the DFW area we have this DFW Blogs meta group. Inside that meta-group are several other groups. When I talk about groups I mean a collection of people who primarily read and comment on each other’s blogs. I’m reticent to give examples of groupage in the DFW Blogs arena as it’s based on my informal observations rather than real data. However, I’m sure most DFW blogger types have observed the groups that have formed.

This is similar to what I talked about back on November 1st about grouping people according to interests. It’s more direct though. I’m not talking about passive implications gleaned from word histograms. It’s about observing the explicit activities of reading and commenting. I imagine both ideas would combine well though. I can use the reading/comment observations to discover the groups, then use the word histograms to figure out the best cross-pollination opportunities.