I've been wondering what types of person to person relationships can be extracted or inferred by analyzing weblog output. My first thought was that there are the explicit relationships that people create in blog rolls and link logs. Then there are the inferred relationships that emerge through audience comments, trackbacks, reverse lookups (Google, Waypath, etc.) and indexes. The diagram below illustrates this initial musing on types of extractable relationships. I break up relationship finding into the explicitly self-identified or knowable and the emergent and discoverable (by machine) relationships. The middle part of the diagram identifies relationship types and the right identifies technologies that might be used to express/display these relationships.

The explicit relationships in blog rolls can be expressed via something like FOAF. Tools like ecademy already allow you to view the person associations you've made via a FOAF file. Hopefully, the serious vendors of social software would offer something similar as well. The observable or emergent relationships, I think, are more in the realm of social software because much of the work there is extraction and analysis. After thinking about this more, I started to wonder if inferring relationships by parsing and extracting the links found in blog entries and link rolls should really be in the observable/emergent box. I think they probably should, so I'm revising my illustration as we go here. To me, this is possibly one of the best types of relationship creation. As patterns of repeatedly pointing to the same person/resource emerge, the strength of relationship is implied. As an aside, this is making me thing of relationship diagrams where strength of relationship is visualized by width of lines. By, that's a digression.

So re-illustrating where I am with this now, I get something like below (also realized I missed including XFN as a supporting standard):

I think this better expresses the difference between explicit author created relationships and observable machine-inferred relationships. This is very high level and something seems to be missing here, but I'm not quite sure what it is.

Are there other types of relationships that I've missed in the center? What other types of technologies or specific applications that should be listed in the right column? This is admittedly very high level -- I'm using it to help communicate some ideas at work -- but at some point programmers will be interested in discussing the specific technologies that are appropriate for these types of expression.