Professor Konstan and CS&E Alums Receive Award for Classic Paper at AAAI

konstan-2010-photo.jpg
February 8, 2017

CS&E alumni Al Borchers, Nathaniel Good, Ben Schafer, Badrul Sawar, and Jon Herlocker, along with Professors Joe Konstan and John Riedl (posthumous), received the 2017 Classic Paper Honorable Mention distinction at the Association for the Advancement of Artificial Intelligence’s (AAAI) annual conference. They were honored for their paper, “Combining Collaborative Filtering with Personal Agents for Better Recommendations.”

Their paper introduced and evaluated a key improvement to recommender systems, which are systems that help users navigate and select among the vast quantities of products or information available on the internet.  Specifically, their work provided a means to combine content-based information filtering with collaborative filtering. Information filtering builds a model of a user's preference in terms of content attributes. For example, a Netflix user might have a preference showing she likes comedy, romance, and the actor Woody Allen. This model is then used to evaluate candidate movies to recommend to that user. Collaborative filtering, by contrast, identifies a community of other users with similar tastes.  For example, that same Netflix user may have 200 other users whose tastes are most similar to hers.  The opinions of this neighborhood of similar users are combined to recommend movies for our Netflix user.

This classic paper showed that information filtering preferences can be merged into collaborative filtering by treating each user's content preferences as a synthetic user—a "filterbot" that rates all the movies in the system according to content attributes.  It found that mixing filterbots and real users together provided a better set of recommendations than either method alone. The combination of content- and collaborative-filtering techniques is now a standard part of recommender systems as practiced across commerce and the Internet.

AAAI-17 was held in San Francisco, California and it aims to promote research in artificial intelligence (AI) and scientific exchange among AI researchers, practitioners, scientists, and engineers in affiliated disciplines

Please join CS&E in congratulating the entire team on receiving this award.

Category: 
Alumni News
Best Paper Awards
CS News