First and foremost, happy shouts out to Josh and Justin. The former became gainful in a job that will keep him in the Midwest (when not flying about), and the latter is just a super really-cool guy. And oh, he won top honors for the student poster at NetSci 2006. Even more amazing an accomplishment since his poster had to speak for itself for about 30 minutes Monday, thanks to a suprise early start to the poster session. Also, Erik gets to code! For the whole summer! As all the employed kids are saying these days, “Luck-yyy.”
Another big shout out to me, too. OUCH! Apparently the act of taking some 61 pages of session notes over the course of a week gave me tennis elbow. It is actually a relief to type. At some point, it might make a nice reference, but I think the main reason I wrote so much was because 90% of this seemed new to me. Unlike Justin — who can now talk the talk like a real Net Scientist — I’ve spent my time trying to pass. I really don’t know an Eigenvector from an Eigenmann, but now I’ve got the makings of a very nice concept map of citations, terms, theories and people. There has been almost no time or energy left at the end of the long days to do much, but NetSci was well worth the resources I spent on it.
In the world of Net Scientry, Albert-László Barabási is like baseball’s Alex Rodriguez. I picture Barabási telling some terminal kid in a hospital that his next correlation of a betweeness centrality eigenvalue thingy will be just for him. Not only did he write a rockin’ book (“Linked”) that makes it all accessible, but he’s credited with taking the first steps to bring this multi-disciplinary bunch together to study networks. I admit to being pretty excited about meeting him. The thought got me out of bed this morning, and I was visualizing the moment passing a fellow conference go-er on the stairs up to the lecture hall.
When I arrived, there was this elderly man sitting at the front table, white hair and beard and sporting all of the signs of a real physicist. The IU people (and others) were being very deferential to him, showering him with respect and attention. I was just about to go thank him for writing his great book, when the guy I passed on the stairway — more George Clooney than Neil Bohr — entered and sat next to him. THAT was Barabási. My mental picture trashed, I never recoverd to chat with him.
The next-to-last day’s highlights:
- Network biology / Albert-László Barabási (Notre Dame) — Great speaker, and nice slides (something of a rarity in this community … Josh! do something! Quick!). Barabási explained why physicists are interested in biology and why we can no longer think of there being a “cancer gene” that causes disease. Instead, there is a dual network of genomes connected to the origins of disease and a disease network connected through shared genes. The really neat thing was that he explained the relationship between the placement of genes and how widespread a resulting disease would be in the body. Peripheral position are where disease genes reside.
- Social networks and capital formation / Jeff Johnson (East Carolina) — One of my favorites from NetSci, Jeff went long (but all the morning speakers did) but had more interesting things of relevance. The pressure to rush to an end combined with text-heavy slides muddled the second study, but the first one (studying perceptions of meat in a Midwestern town) demonstrated the effects of culture on perception of meat. The upper class participants saw food as instrumental, rather than just functional, and thus categorized meat as a process. The lower-middle class group didn’t typically socialize around food. This was all interpreted through network data and a card sort of different meats. What I like most about Jeff’s work is how much he explains the methodology in a way helpful to any designers in the audience.
- Dynamic communication networks during extreme events / Carter Butts (UC Irvine) — My appreciation for this guy grew daily. I still don’t understand most of what he talks about, but it is clear Carter is brilliant and very much in tune with all things equation. I also appreciated his ability to ask very relevant questions when others spoke. He exuded that confidence again Wednesday morning, though much of his talk was still big math. The topic, though, was interesting: examining the transcripts of radio communication during the 9/11 disaster in New York. He was trying to see if the dynamic nature of this network would indicate what kinds of strategies people used to communicate with each other.
- Networked derived visualization of U.S. Supreme Court decisions / Peter Hook (Indiana) — This was a well-done talk examining a recent 10-year period where all 9 Supreme Court justices were together. The impetus for this study was the low level of useful information coming out of the graphs and tables the media was using to show how pivotal Sandra Day O’Connor was to judicial decisions. Hook didn’t really hit upon a winning visualization, but the network analysis showed some interesting associations, particularly with how the justices changed from issue to issue.
- Network ranking method for college football / Juyong Park (Notre Dame … formerly Michigan) — This was the workshop poster I was looking forward to seeing that never showed. Park had a great presentation that was very clear in every aspect what he was doing and the motivation for doing it. Basically, instead of the convoluted formulas for determining BCS pecking order for bowl games, he proposed a system of using multiple depths of indirect wins and losses (“My team beat the team that beat you, so we’re better”). Looking back over the available history of the data, he claimed to be about 80% accurate … including this year’s champion Texas team. I’m glad there were a couple sports-centered uses of network analysis at this conference.
- Automatic extraction of concealed relations from email logs / Nishith Pathak (Minnesota) — I got a little lost here, but the main point is that there was evidence through the network of email that the Enron bigwigs had some email exchanges they concealed from others.
- Layered complex networks / Maciej Kurant (EPFL/Switzerland) — What was neat/useful about this is the idea that multiple layers not only exist simultaneously for transportation networks (i.e. trains running on train tracks) but that there may be useful information that could be gleaned from combining their analysis. Since I’m interested in understanding the actor and information networks that arise in forums, this might have some future relevance.
- Rich-club ordering in complex networks / Vittoria Colizza (Indiana) — A very well-done talk by one of our own explained the idea of “rich club” networks where the nodes with the highest degrees are also well connected with each other. Scientific collaboration and air transportation networks are examples of this. This, too, has relevance to forum studies.
One more morning session for me tomorrow.
