Acquiring Unobtrusive Relevance Feedback through Eye-Tracking in Ambient Recommender Systems

Acquiring relevant information to keep user’s preferences up-to-date is crucial in recommender systems in order to close the cycle of recommendations. Ambient Intelligence is a suitable approach for non-intrusively closing the loop in recommender systems using ambient eye-trackers. We combine a method for acquiring relevance feedback through eye-tracking with the functionalities of an extractor agent. We describe the results of experiments conducted in a recommender system to obtain implicit feedback using eye fixations. Finally, we obtain a ranking of user’s most relevant preferences and behaviours. Keywords. User Modelling, Ambient Recommender Systems, Implicit Relevance Feedback, Eye-Tracking

• Type: Conference
• Year: 2005
• Where: CCIA
• Authors: Gustavo González, Beatriz López, Cecilio Angulo, Josep Lluís de la Rosa
• Keywords: Ambient Recommender Systems, CCIA 2005, Recommender Systems, Relevance Feedback