dynamic information filtering

up
patrick baudisch
  projects
       dynamic IF
dynamic IF
ph.d. thesis
defense talk
publications
tv-scout
paintable UIs
visitor counter
schematic diagram showin the profile updating mechanism of the query set architecture The QuerySet Architecture computes output rankings in two steps. First, the relevance of objects with respect to the user's individual interests is computed (top), then all these relevance ratings are aggregated into a single output rating (middle).

The QuerySet Architecture is an information filtering architecture optimized for frequently changing user interests.  QuerySet systems can track gradual interest changes by learning from relevance feedback. In addition to that, users are provided with direct access to their user profiles (e.g. using paintable interfaces) to allow them to manually handle major profile inaccuracies, as they occur during profile initialization and after major interest changes. The validity of the QuerySet Architecture was demonstrated at the example of the TV program recommendation system TV Scout.