“Searching Public Discourse”
by Maarten de Rijke
Data has become big. But the volume is not the only aspect that has changed. We now have massive volumes of data from open sources that can inform us about human behavior. Millions of people share views and reports on experiences online, either with a specific circle of followers or with no one in particular (i.e., the entire planet). Often covering extended periods, data culled from social media platforms can tell us a lot about people's lives. But what can it tell us? We share experiences and perspectives -- but there's more: our transactions are logged, our cars and homes monitor our movements and activities, a preferences expressed through "likes" are aggregated, body and brain signals are becoming a commodity. An important challenge for computer science is to design search and analysis methods based on self-learning algorithms that exploit a multitude of signals of this type to improve their functioning. We have reached a point where the boundaries between automated information-processing systems and their human users are blurring. Increasingly, such systems are becoming part of the online discourses in which we all engage. The talk will be driven by a number of examples plus a look at algorithms for search engines that learn to adapt their rankings from interactions with users.