REFER Seminar
Computer Science Department
Worcester Polytechnic Institute

Date: Nov. 8, 1999

Speakers: Profs. Dave Brown & Mark Claypool, WPI CS Dept.

Title: "Implicit Interest Indicators" or "I's Cube"


ABSTRACT

   Recommendation systems are often intrusive, as in order to get an
   explicit measure of the user's interest in an item (such as a web
   page), the user has to type in some sort of score or select an
   interest value from a scale.  As computers should serve people as
   much as possible, and not vice versa, it would be ideal if the
   system were able to accurately detect and measure interest without
   the user being aware of it or having to act differently, That would
   allow accurate recommendations to be given `as if by magic'.  Hence
   there is a need for interest to be determined by aspects of the
   user's behavior.  As they indicate interest, and are not provided
   explicitly by the user, we use the term "implicit interest
   indicators".  In this presentation we describe our ideas for some
   research that will study and evaluate such indicators.