Content-Independent Task-Focused Recommendation

  • Authors:
  • Jonathan L. Herlocker;Joseph A. Konstan

  • Affiliations:
  • -;-

  • Venue:
  • IEEE Internet Computing
  • Year:
  • 2001

Quantified Score

Hi-index 0.01

Visualization

Abstract

Recommender systems, also known as collaborative filtering systems, predict a user's current interest in unseen items based on ratings or recommendations from other people. Users are typically asked to rate items (such as movies) that they have already seen or experienced. The recommender system then matches each user up with other people, known as neighbors, who have given similar ratings and recommends items that the user's neighbors have rated highly. Such recommender systems assume that a user's interest is based solely on historical ratings data and is independent of the user's current task. In reality, the user's current context or task greatly affects the value of a recommendation. This article presents a new approach that provides task-focused recommendation that is independent of the type of content being recommended. The authors have implemented the proposed system on a movie recommendation site and validated it with empirical results from user studies.