Matrix Factorization Techniques for Recommender Systems

  • Authors:
  • Yehuda Koren;Robert Bell;Chris Volinsky

  • Affiliations:
  • Yahoo Research;AT&TLabs;AT&TLabs

  • Venue:
  • Computer
  • Year:
  • 2009

Quantified Score

Hi-index 4.16

Visualization

Abstract

As the Netflix Prize competition has demonstrated, matrix factorization models are superior to classic nearest-neighbor techniques for producing product recommendations, allowing the incorporation of additional information such as implicit feedback, temporal effects, and confidence levels.