Web Science: A Royal Society Meeting (day 1)

notes on the first day of the royal society event web science. featured: lazlo barabasi, lord may, jon kleinberg, and david karger ;-)

barabasi ‘s talk titled “the structure of the web”  was already mentioned on this blog as it was given in the past in several occasions, eg, “network scince” & sigir

lord may gave a talk on how ecosystems, infectious diseases, IT networks and financial systems are all linked (“Understanding what happens when you disrupt networks is important in many ways”, he said  (we agree, see, eg, bank collapses)). lord may is a good speaker and a serious comedian ;-) interestingly, he mentioned a general caveat/problem for this kind of studies: we can often only study sub-nets of the networks, which only represent the networks as a whole in restrictive networks. also there are a lot of dynamic elements that cannot be considered [Nature 451 893-895 2008]. we should all remember that ;-)

Jon Kleinberg gave a fantastic talk – “understanding social and information networks”. kleinberg started off with a great statement: “science advances when we make the invisible visible”, ie, when we make the invisible (our social interactions) visible (via flickr and facebook and the rest). kleinberg briefly covered a paper in which they try to extract from flickr the most significant (salient) pics for geographic locations. why is that useful? well, one could think of using flickr  photos (which represent crowd-sourced endorsements of locations) to automatically build  travel guide ;-) kleinberg then covered few papers that try to answer three fundamental questions:

  • Q1. what is the probability you form new friendships or engage in new actives based on behaviours of existing friends? He showed that the probabilty of doing something (e.g. joining online group) when 2 of your friends are doing it is more than double if only one was. for more work on this, please refer to  backstroke 2006(joining livejournal group) , cranial 2008 (editing wikipedia), backstroke 2006
  • Q2. why are you similar to your friends? because they influence you or because you seek out people who are already similar?
  • Q3. how do you evaluate other people? kleinberg mentioned a recent paper in CHI that tries to infer both positive and negative links of social networks (i can imagine some connections with sentiment analysis in twitter)

The main point of Kleinberg’s talk was that  social networks – implicit for millennia – are now being recorder at arbitrary resolution and embedded in the web. As such, we have the opportunity to understand why social processes turn out the way they do, with insights for design, analysis, and prediction! Great great talk!

currently, data on the web is largely published by private and public organisations (e.g., data.gov), and the general public doesn’t do it (more specifically, those who aren’t tech-savvy are practically excluded). karger of mit argues that APIs are a barrier to data reuse (you need to learn how to read/use API first) and, as such, his team is working towards making publishing data on the web as easy as it is to publish a webpage. more here http://www.simile-widgets.org/exhibit/. he also showed a wordpress blog plugin that makes it easy to publish data and  the exhibits generated by the public. fantastic work, which will make the

idea of semantic web practical. cheers!

this is an incomplete list of talks. more here

Comments are closed.