ICWSM Recap

I'm back home from my trip to Seattle, and the International Conference on Weblogs and Social Media.  I had a great time, despite having my return flight canceled.  I liked Seattle, at least what little bits I got to visit. And I had a fun impromptu trip to Safeco Field to see the Mariners play the Rangers (Mariners won). 

This is the first academic conference I have attended in one of my core research interests.  It was pretty exciting to be able to interact with other people in academia and industry with similar interests, especially since there seems to be so few people at Iowa doing research in this area.  More about the conference after the break...(technical content ahead).

Tags: icwsm, icwsm2008

It seems like a people have been writing about the conference, so I won't do a day-by-day recap.  But these are the ideas and themes that stuck with me... Who are the social media researchers? Everyone, it seems, has an interest or a stake in this field.  Academic institutions around the world, big and small, companies big and small. This is definitely a young field: most of the attendees identified themselves as PhD student.  The nice thing about this field being so young (in the prescience phase, if you read Kuhn) is that there are many opportunities to make a contribution.  The downside is that it is difficult to make any real comparisons between works.  What do we know, or agree upon?

• Small world network properties in the graph structures we're studying
• Prevalence of power law distributions in network links
• Blog/social networks tend to be pretty sparsely connected

What don't we know, or agree upon? Just about everything else.  As I mentioned before, we're in a prescience phase, in which there is no real central paradigm.

• What are the important dimensions or classifications of social media?  What exactly is the definition of social media? Marc Smith of Microsoft Research had some interesting thoughts on this (I hope someone posts a copy of his keynote slides...Matthew?)
• What are the standard metrics for social graph structures? Some people used Entropy or mutual information, some used F-measure, some used a correlation coefficient. 
• There were many correlational studies, which makes it difficult to understand or assign causality to the phenomenon we're studying. 
• What attributes of online profiles are important for which application or environment.
• Most importantly, SO WHAT? We're doing lots of work looking at different things...why?  Because the data is there?  Besides commercial interests, what public needs can we fill with this work?

What problems are currently important or interesting?

• Identifying communities, and key influencers within a social network (although Dave Sifry argued in his invited talk that may not be as important.
• Sentiment analysis of blog networks: what do the community of bloggers feel about a particular topic or item.  Obvious applications in politics and marketing.
• Characterizing individual behaviors in particular communities.  Microsoft Live Labs had some nice visualizations that helped them identify different types of user behavior in social networks.
• Scalable algorithms and techniques.  Need to scale in time, disk space.  This was highlighted as a disconnect between academia and industry.
• Interoperability and exchange of data between networks (highlighted by Brad Fitzpatrick in his keynote talk).
• Spam and deception detection.  There are at least 2 universities (UMBC, Texas A&M) doing work on spam and deception detection.

What are the common tools, methods, or techniques used?

• People are writing lots of custom webcrawlers/screen scrappers. 
• Machine learning techniques: clustering, classification
• Matrix math/manipulation.  I found I need to learn more about tensors
• Language models.  Lots of people used a simple 'Bag of Words' approach to understanding the text in blogs.  Fewer used n-grams.  There was one paper that discussed using more advanced linguistic models to perform sentiment analysis.

So I learned lots about the field, and met lots of people that were interested in the same sorts of things that I am.  More about how my research will change as a result will be in a future post.

Comments

[alison lewin]

jerry ! sarah ! it's been so long...over two years infact...I think about you all the time and cherish the beautiful photo I eventually got of antalya...I say eventually because I had vacated the premises at Church road and it was a few months later when the post was finally redirected to me and ...and in there was your beautiful baby daughter.... I have found your page and have no idea if this will reach you but I'm hoping that you get the message somehow that you are both in my thoughts.....
I'm still catering and still acting...what a combination !
lots and lots of love...hopefully I will hear from you soon.
Lots of love
Alison

[Josiah]

This sounds fascinating. Perhaps I should consider a computer science major...