I found last week’s AAAI Symposium on Computational Approaches to Analyzing Weblogs to be a stimulating mix of theory, practice and speculation. Of the eight symposia in the series this Spring, this was most popular, with over a quarter of the 300 attendees. The symposium’s blog failed to ignite, probably for logistical reasons. But maybe to some degree it’s a case of “those who can, blog; those who can’t, do research on blogs”. Here were some of the highpoints for me.

• Geoffrey Nunberg gave a literate (surprise!) and interesting talk on blogs as a medium for “the public”. His hour talk ran 30 minutes over it’s one hour slot and into the lunch break, and no one in the audience stirred.
• The industry panel on Technologies to Understand it Now and Gain Insight in the Future was interesting and generally informative, but some of the responses to questions were quite guarded. One thing I learned from it was the concept of consumer-generated media which includes blogs, of course. (See Mathew Hurst’s post on social media.) Another idea I found intriguing was the corporate use of blogs, such as the internal blog used by Intel CEO Paul Otellini to communicate and interact with Intel employees.
• There were several disappointments in the panel, though. When asked what what problems researchers could help solve, nothing significant was mentioned. Maybe they panelists were keeping their cards close to their chest. When asked about the possible role of the semantic web, e.g., in support of structured blogging, data blogs or more metadata, no one seemed optimistic or even interested.
• I was surprised how many of the papers were based entirely on data from LiveJournal blogs. While LJ makes its data and metadata easy to get, it represents a narrow view of the Blogosphere.
• That said, I found the paper, demo and presentation on MoodViews fascinating. MoodViews is a collection of tools for tracking the stream of mood-annotated text made available by LiveJournal.
• Matthew Hurst and Natalie Glance of Intelliseek/BuzzMetrics gave three (!) presentations that were all very interesting and useful.
• 

  UMBC’s own Pranam Kolai did a great job of talking about our work on splog identification. The environment wasn’t good for pictures, but I got one shown on the right. Pranam had the brilliant idea of wearing a UMBC sweatshirt for his talk. I think we should make this our lab’s standard dress for presentations.

• There were lots of other good talks and papers. Hopefully, these will be available on the CAAW site before too long.

Categories: Uncategorized, , Related posts: • AAAI Sympoium on Computational Approaches to Analyzing Weblogs;  • Computational Approaches to Analyzing Weblogs;  • SVMs for the Blogosphere: Blog Identification and Splog Detection;   Trackback: link, Comments: