Your SlideShare is downloading. ×
Aastha jain youtube 05 sept 2012.pptx
Upcoming SlideShare
Loading in...5
×

Thanks for flagging this SlideShare!

Oops! An error has occurred.

×

Introducing the official SlideShare app

Stunning, full-screen experience for iPhone and Android

Text the download link to your phone

Standard text messaging rates apply

Aastha jain youtube 05 sept 2012.pptx

200
views

Published on

Published in: Business, Technology

0 Comments
0 Likes
Statistics
Notes
  • Be the first to comment

  • Be the first to like this

No Downloads
Views
Total Views
200
On Slideshare
0
From Embeds
0
Number of Embeds
0
Actions
Shares
0
Downloads
1
Comments
0
Likes
0
Embeds 0
No embeds

Report content
Flagged as inappropriate Flag as inappropriate
Flag as inappropriate

Select your reason for flagging this presentation as inappropriate.

Cancel
No notes for slide

Transcript

  • 1. Multi domain recommendation systemMOVIES BOOKS MUSIC
  • 2. Multi domain recommendation systemMOVIES BOOKS MUSIC
  • 3. Multi domain recommendation systemMOVIES BOOKS MUSIC
  • 4. Multi domain recommendation systemMOVIES BOOKS MUSIC
  • 5. Multi domain recommendation systemMOVIES BOOKS MUSIC
  • 6. Multi domain recommendation systemMOVIES BOOKS MUSIC
  • 7. Multi domain recommendation systemMOVIES BOOKS MUSIC
  • 8. Multi domain recommendation systemMOVIES BOOKS MUSIC
  • 9. Multi domain recommendation systemMOVIES BOOKS MUSIC
  • 10. User characterization by item preferences Movies BooksJohn likes:Mary likes: Recommendations:Tom likes: ??
  • 11. User characterization by item preferences Movies BooksJohn likes:Mary likes: Recommendations (domain-specific): Popular books suggestedTom likes: since no prior information is available
  • 12. User characterization by item preferences Movies BooksJohn likes:Mary likes: Recommendations (multi-domain): Based on Tom’s movie preferences, and otherTom likes: users’ cross-domain information
  • 13. User characterization by item preferences Movies BooksJohn likes:Mary likes:Tom likes:
  • 14. User characterization by item preferences Movies BooksJohn likes:Mary likes: Recommendations (multi-domain):Tom likes:
  • 15. User characterization by item preferences Movies BooksJohn likes:Mary likes: Recommendations (multi-domain):Tom likes:
  • 16. New items or sparsely rated items • New movie • Very few user ratings • Cannot be correctly classified and recommended • Use meta information • Easier to identify similar items
  • 17. New items or sparsely rated itemsTom likes:
  • 18. New items or sparsely rated itemsTom likes: Recommendations: OOPS!
  • 19. New items or sparsely rated itemsTom likes: Recommendations: OK!
  • 20. # items rated by user Average squared error User Domain Specific Recommendation# items rated by user Average squared error User Multi Domain Recommendation
  • 21. Domain Specific Recommendation Multi Domain RecommendationAverage squared error Average squared error 62.56% decrease User in root mean User squared error# items rated by user # items rated by user
  • 22. Quantitative Improvement Method Percentage Books Music Video accuracy (220,000 (80,000 (22,000 ratings) ratings) ratings)Collaborative 49.4 50.21 48.1 47.7 Filtering (domain specific) Collaborative 63.2 65.67 60.22 57.34 Filtering(multi-domain) Hybrid 76.4 77.13 75.11 73.87(Multi-domainCollaborative +content based)
  • 23. Aastha Jain