Just sent in an abstract for a paper Jia Lin & I are doing for the Sunbelt social networking conference.
What Do Network Measures Predict on the Web?
Social network approaches have been used extensively to help to make sense of hyperlink networks on the World Wide Web. Measures of prestige drive, for example, Google’s PageRank system. While there is no lack of argument suggesting that such measures provide a useful indication of the “quality” of a particular search or page, there is little research demonstrating that such measures effectively predict either the popularity of a page or the likelihood that it will be linked to. We examine a subsection of a large collaborative “wiki” website and use several network derived measures to determine the best way to predict both popularity (“hits”) of particular pages and the likelihood that they will receive new hyperlinks. While the collaborative nature of the site makes it somewhat unusual, we argue that this represents some indication of the power of such measures in describing the larger Web.
I’d planned on also presenting the work on co-posting networks I’ve been slaving away on for the JCMC special issue, but after a double-presentation at AIR, I decided to hold off and try to budget my time more reasonably. Besides, if it gets published in JCMC, I’ll end up getting better (or at least more) feedback that way anyway.