Not only has the Internet revolutionized human life, it has also revolutionized the study of human life. Whereas people’s pre-Web, offline behaviors went largely undocumented, nearly every online action is recorded in log files. This way, human behavior can now in principle be analyzed in unprecedented detail. In my research, I have been working at the intersection of data mining, machine learning, and natural language processing to convert raw log data into meaningful insights on a number of human behaviors, such as navigation in complex networks, Wikipedia editing, linguistic change in online communities, and food intake.
The insights we gain through our analyses are useful as they allow us to turn vast amounts of behavioral log data into actionable policies. For instance, we showed that, by analyzing the queries with which users search for online cooking recipes, we can understand at a large and cheap scale what foods are popular where at what times. This information might in turn be used to craft better public-health campaigns, targeted specifically at sub populations at risk of eating unhealthy foods. As another example, we conducted a longitudinal study of how the language of online-community members changes over the years and demonstrated that users about to leave the community fail to keep adapting their personal language to community-wide linguistic novelties. This lets us predict whether a user will leave a community, an important task for many websites that depend on large active user bases.
The Facebook Fellowship gave me the freedom to choose more autonomously what to work on, which in my opinion is crucial for creativity. I also deeply enjoyed the benefits of the academic-travel allowance, which allowed me to attend several conferences that I would otherwise not have been able to attend, in places I had never been to before. Thank you very much, Facebook!
In my next project, I am planning to understand better how the online media world (mainstream media outlets, blogs, and social networks) reacts to events happening in the world: what events receive broad coverage, what others are largely ignored, and how does this help shape public opinion as measured in blogs and social networks?