Your SlideShare is downloading. ×
  • Like
Jon Sanders on Collaborative Filters at SXSW
Upcoming SlideShare
Loading in...5

Thanks for flagging this SlideShare!

Oops! An error has occurred.


Now you can save presentations on your phone or tablet

Available for both IPhone and Android

Text the download link to your phone

Standard text messaging rates apply

Jon Sanders on Collaborative Filters at SXSW


Jon Sanders of Netflix presenting "Collaborative Filters: The Evolution of Recommendation Engines" at SXSW Interactive, March 14 2009

Jon Sanders of Netflix presenting "Collaborative Filters: The Evolution of Recommendation Engines" at SXSW Interactive, March 14 2009

Published in Technology , Business
  • Full Name Full Name Comment goes here.
    Are you sure you want to
    Your message goes here
    Be the first to comment
No Downloads


Total Views
On SlideShare
From Embeds
Number of Embeds



Embeds 0

No embeds

Report content

Flagged as inappropriate Flag as inappropriate
Flag as inappropriate

Select your reason for flagging this presentation as inappropriate.

    No notes for slide


  • 1. Personalizing Netflix A brief history Jon Sanders Recommendation Systems Engineering Netflix Los Gatos, CA jsanders @
  • 2. Fun facts about Netflix World’s largest online movie rental service #1 in customer satisfaction Founded 1997 Video rental companies (Consumer Reports) Online retail (ForeSee) With more than… On an average day 10M subscribers, $1B revenue 2M DVDs shipped 100K DVD titles, 50 distribution centers 2M movie ratings received 12K streaming movies & TV episodes 1.5B minutes streamed to 1M Xbox360’s 2B movie ratings 60% of movies selected based on personalized recommendations Connecting people with movies they’ll love
  • 3. In the beginning… Everyone sees the same site
  • 4. Evolve methodically
  • 5. The rating widget •  Ask about & predict movie Enjoyment •  User-similarity collaborative filter •  Recommendations fuel discovery
  • 6. Score & sort any movie Combine popularity & enjoyment prediction
  • 7. Tune recommendations •  Movie-similarity collaborative filter •  K-nearest-neighbor algorithm •  More credible connections
  • 8. Interest-based discovery Metadata connections: actor, director, genre, …
  • 9. Ask about Interest Moderate prominence of catalog areas
  • 10. Ask other people Community offers decision support
  • 11. Explain why Build trust with reflected evidence
  • 12. $1M Netflix Prize •  Improve accuracy of Enjoyment predictions –  100M ratings –  Achieve 10% better than Netflix RMSE •  Innovative, engaged research community •  Highly relevant results –  Global and time-based corrections –  SVD, RBM models –  Blending predictors
  • 13. A website for each subscriber
  • 14. Evolution continues •  Tailor with more metadata, implicit data •  Streaming-specific personalization •  Collaborative Filtering is a component of personalization •  People want to drive, not be led •  Offer discovery, focus and decision support
  • 15. Links • • • • •