Milwaukee.
The final speaker in this session at AoIR 2009 is Anatoliy Gruzd, whose focus is on the communities of blog readers, and how such communities of people discussing shared issues across different blogs may be discovered automatically - that is, how the social networks connecting them may be identified. This is important not least because of the massive growth in online information - we need to develop better tools to extract salient material from this overload of content, and to do so, knowing the social context is paramount.
Other such networks include email, forum, blog, co-author and co-citation, and friends networks, and in each of these cases, there are now efforts to identify the social networks connecting individuals. This has commercial applications, but also enables researchers to track the growth and decline dynamics of social communities.
To begin with, though, Anatoliy's study focusses on a specific blog and its commenters - can it support the development of a social media community? He began by identifying personal names in comments, and from this extracting personal connections (if a commenter refers to someone else, we can infer a connection). Names are used to identify an addressee and attract their attention, and to maintain and reinforce social relationships; they are one of the few textual carriers of personal identity. Anatoliy developed the Internet Community Text Analyser Website (ah! I've been meaning to look at this!) to do this, and this is now also available to other researchers.
The blog Anatoliy has focussed on is that of Garth Turner, former Canadian politician turned author and real estate guru - can this be described as a virtual settlement; is there evidence of a sense of community here? He used the Community Text Analyser to identify the changes in social networks on the site over time; from October 2008 to January 2009, there is an obvious change in the network structure that emerges: when at first, there was a central individual acting as a central connection point between regular commenters, this function is now much more distributed.
Overall, he found a sustained membership on the site (44% of users were present during both timeframes), with highly interactive discussions frequently referencing other users, sustained by interpersonal (and sometimes off-topic) interactions, and enabled by self-moderation. So, it seems like social networks can be discovered in this way, and the study shows that it is possible for viable social network communities to emerge amongst the commenters on a single blog.
