Central to the paper is the notion of ad-hoc topic communities. Suppose that you are worried about ice cap melting (the topic), and want to find people who have something to say on this. The paper describes an approach on how an ad-hoc topic community might be derived from a set of blog posts on ice cap melting and related topics such as global warming.
The title of the paper is ``A socio-technical approach for topic community member selection''. One of the reviews commented that communities cannot be constructed from a topic, i.e. ``topic community'' is not a sensible term. S/he suggested that ``communities of interest'' might be a better term. It is difficult to agree or disagree here. The underlying point of the reviewer is undoubtedly that to be part of a community some traceable connection to other community members is required, and ideally the connection is reciprocal. On the other hand, the term community is used frequently for a group of people that have something in common without necessarily knowing each other (e.g. the Muslim community).
After submitting the paper and blogging about the process of writing it (see From a BlogPulse search to a corpus) Natalie Glance provided us with direct access to the BlogPulse database. And the direct access makes it possible to verify whether potential ``community of interest'' members also link to each other, hopefully addressing the issue raised by the thoughtful reviewer. The final version of the paper is due in a month.