KEYNOTE SPEAKER

Attentive Computing - Employing Gaze Data for Anticipating a User's Thoughts

Prof. Andreas Dengel
DFKI & University of Kaiserslautern, Germany

Abstract

Intelligent systems monitoring their users and observing their work are becoming more and more important. Correlations between human actions and document contents are essential for building "electronic information butlers" that are able to anticipate what may be required in a given context. Due to the limitations of today’s computer interfaces in user observation, we have to consider new approaches for connecting human thoughts to relevant information.
In this talk, I would like to propose a new but highly promising research field where gaze data collected during reading is used to measure attention while reading. Fixation points and saccades are taken to determine reading behavior and for calculating contextual relevance of content while working with documents. The findings are employed for a new relevance measure that improves retrieval as well as classification significantly. Furthermore new options for human-computer-interaction arise, i.e. a text may be enriched by invisible mark-ups augmenting a simple alphanumeric document with a virtually multimedia world behind the text. In my talk, first, I will give some insights into our approach for combining the brain and computer-resident information via the eyes and then show the new options and potentials of this emerging research field for application such as learning, classification or search. I will conclude my talk by giving an outlook to other innovative fields that may be influenced by Attentive Computing technologies.