MarCol: A market-based recommender system Academic Article uri icon

abstract

  • Collaborative information-filtering systems recommend relevant items to users on the basis of their common interests. The users express their interests by leaving relevance feedback on items. The system's ability to learn user preferences and predict accurate recommendations depends on the number of judgments the user provides. However, users tend to "free-ride," consuming other users' judgments without providing their own. To solve this problem, systems should offer users incentives for providing judgments. A new market-based model for pricing judgments aims to motivate users by requiring them to provide judgments before they can receive recommendations. Researchers used MarCol, a market-based collaborative IF system, to conduct experiments examining the model's effect on user feedback provision, user satisfaction, and recommendation quality. Results show that the model increases feedback and improves recommendation quality. This article is part of a special issue on Recommender Systems.

publication date

  • January 1, 2007