Behavior on the web has become increasingly interactive with the rise of user-created content sites such as flickr, LiveJournal, Facebook, and Wikipedia. These sites capture and provide access to information about the interactions between people and the artifacts they create, opening up rich opportunities for computer scientists and social scientists to come together and answer questions about human behavior--what motivates people to contribute? why do they join groups? how do social connections influence other people?--on a scale never possible before.

Our approach is twofold. First, we collect publicly available data from the communities listed above, using that data both to create models of behavior and to address open questions in fields such as sociology and economics. Work in progress includes understanding how people adopt roles in online communities, understanding the diffusion of new ideas through an online community, and measuring how the relative effects of social influence and one's own interests affect people's future behavior.

Second, we use our understanding and the data to build tools to improve these communities. Wikipedia's SuggestBot tool, reported on in the New Scientist, has helped over 5,000 people on Wikipedia find articles to work on by using an idea called intelligent task routing. It matches people's interests over time with those of other people to find things they might want to help with, and chooses the ones that Wikipedia itself has marked as needing the most contributions.