David Reitter is a Research Fellow in the Adaptive Speech Interfaces group, where he works on multi-modal, natural language based communication.
Recently, David's research has been concerned with the analysis of rhetorical structure in texts. How do people construct an essay? How can a computer automatically identify the important parts of written text to give a concise summary? Machine Learning algorithms increase the coverage of such systems by learning from real-world data. To work well, they need to be augmented with linguistic knowledge about texts.
David is now applying rhetorical structure to the broader domain of multi-media. By using sounds, speech, images and text simultaneously, we can speed up and simplify human/computer and mass media communication. A multi-modal computer could adapt to the user's current situation, as well as to specific needs of disabled or confused users. However, several constraints and factors govern the combined presentation of different material. An underlying theory may provide a framework to handle multi-modal fusion and output. The new techniques are demonstrated in a multi-modal personal information management system, developed in collaboration with several European partners in the FASiL project.
Prior to joining Media Lab Europe, David led the natural language research of a start-up company in Berlin and Bucharest, and worked as freelance journalist for major German radio stations.
David holds a (post-graduate) Diplom in Linguistics from University of Potsdam. He is a PhD candidate at the University College Dublin.
Visit David Reitter's personal web site...