Your SlideShare is downloading. ×
0
Personalizing Tags: A Folksonomy-like Approach for Recommending Movies
Personalizing Tags: A Folksonomy-like Approach for Recommending Movies
Personalizing Tags: A Folksonomy-like Approach for Recommending Movies
Personalizing Tags: A Folksonomy-like Approach for Recommending Movies
Personalizing Tags: A Folksonomy-like Approach for Recommending Movies
Personalizing Tags: A Folksonomy-like Approach for Recommending Movies
Personalizing Tags: A Folksonomy-like Approach for Recommending Movies
Personalizing Tags: A Folksonomy-like Approach for Recommending Movies
Personalizing Tags: A Folksonomy-like Approach for Recommending Movies
Personalizing Tags: A Folksonomy-like Approach for Recommending Movies
Personalizing Tags: A Folksonomy-like Approach for Recommending Movies
Personalizing Tags: A Folksonomy-like Approach for Recommending Movies
Personalizing Tags: A Folksonomy-like Approach for Recommending Movies
Personalizing Tags: A Folksonomy-like Approach for Recommending Movies
Personalizing Tags: A Folksonomy-like Approach for Recommending Movies
Personalizing Tags: A Folksonomy-like Approach for Recommending Movies
Personalizing Tags: A Folksonomy-like Approach for Recommending Movies
Personalizing Tags: A Folksonomy-like Approach for Recommending Movies
Personalizing Tags: A Folksonomy-like Approach for Recommending Movies
Personalizing Tags: A Folksonomy-like Approach for Recommending Movies
Personalizing Tags: A Folksonomy-like Approach for Recommending Movies
Personalizing Tags: A Folksonomy-like Approach for Recommending Movies
Personalizing Tags: A Folksonomy-like Approach for Recommending Movies
Upcoming SlideShare
Loading in...5
×

Thanks for flagging this SlideShare!

Oops! An error has occurred.

×
Saving this for later? Get the SlideShare app to save on your phone or tablet. Read anywhere, anytime – even offline.
Text the download link to your phone
Standard text messaging rates apply

Personalizing Tags: A Folksonomy-like Approach for Recommending Movies

1,289

Published on

Presented at HetRec 2011 …

Presented at HetRec 2011
http://ir.ii.uam.es/hetrec2011/

1 Comment
1 Like
Statistics
Notes
  • Good presentation. I would also like to see directors as a tag.
       Reply 
    Are you sure you want to  Yes  No
    Your message goes here
No Downloads
Views
Total Views
1,289
On Slideshare
0
From Embeds
0
Number of Embeds
2
Actions
Shares
0
Downloads
42
Comments
1
Likes
1
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. Outline Movie Recommendation Folksonomies Our Approach Experiments Conclusion & Future Work Personalizing Tags: A Folksonomy- like Approach for Recommending Movies Alan Said Benjamin Kille Ernesto W. De Luca Sahin Albayrak {alan, kille, deluca, sahin}@dai-lab.de DAI-Lab TU-Berlin HetRec, 2011HetRec2011 :: Said, Kille, De Luca, Albayrak Personalizing Tags 1 / 18
  • 2. Outline Movie Recommendation Folksonomies Our Approach Experiments Conclusion & Future WorkOutline Movie Recommendation Folksonomies Our Approach Experiments Conclusion & Future Work HetRec2011 :: Said, Kille, De Luca, Albayrak Personalizing Tags 2 / 18
  • 3. Outline Movie Recommendation Folksonomies Our Approach Experiments Conclusion & Future WorkAbstract Problem: How to simply use semantic data (tags, genres, etc.) in usage-based collaborative filtering? Aim: To provide a basic model of hybridization without adding algorithmic complexity to a collaborative filtering recommender system. HetRec2011 :: Said, Kille, De Luca, Albayrak Personalizing Tags 3 / 18
  • 4. Outline Movie Recommendation Folksonomies Our Approach Experiments Conclusion & Future WorkMovie Recommendation Traditional approach: Use users’ rating to find nearest neighbors/latent factors/etc. HetRec2011 :: Said, Kille, De Luca, Albayrak Personalizing Tags 4 / 18
  • 5. Outline Movie Recommendation Folksonomies Our Approach Experiments Conclusion & Future WorkMovie Recommendation Traditional approach: Use users’ rating to find nearest neighbors/latent factors/etc. Traditional hybrid approach: Combine two or more parallel algorithms. HetRec2011 :: Said, Kille, De Luca, Albayrak Personalizing Tags 4 / 18
  • 6. Outline Movie Recommendation Folksonomies Our Approach Experiments Conclusion & Future WorkMovie Recommendation Traditional approach: Use users’ rating to find nearest neighbors/latent factors/etc. Traditional hybrid approach: Combine two or more parallel algorithms. Our Approach: Combine several data sources prior to recommendation process - uses one algorithm. Keep implementational effort low - allow easy implementation in existing system. HetRec2011 :: Said, Kille, De Luca, Albayrak Personalizing Tags 4 / 18
  • 7. Outline Movie Recommendation Folksonomies Our Approach Experiments Conclusion & Future WorkDefinition Definition: the result of personal free tagging of information and objects . . . for ones own retrieval [Vander Wal, 2004] Tags offer a short content-related description of items to which they are assigned. HetRec2011 :: Said, Kille, De Luca, Albayrak Personalizing Tags 5 / 18
  • 8. Outline Movie Recommendation Folksonomies Our Approach Experiments Conclusion & Future WorkRelevance? So..how is this relevant to movie recommendation? HetRec2011 :: Said, Kille, De Luca, Albayrak Personalizing Tags 6 / 18
  • 9. Outline Movie Recommendation Folksonomies Our Approach Experiments Conclusion & Future WorkRelevance? Our movies have tags, e.g. categorized with tags from five cate- gories: Moods Places Times Intended Audiences Plots HetRec2011 :: Said, Kille, De Luca, Albayrak Personalizing Tags 7 / 18
  • 10. Outline Movie Recommendation Folksonomies Our Approach Experiments Conclusion & Future WorkRelevance? HetRec2011 :: Said, Kille, De Luca, Albayrak Personalizing Tags 7 / 18
  • 11. Outline Movie Recommendation Folksonomies Our Approach Experiments Conclusion & Future WorkNot quite a folksonomy We have a problem: Tags are not personalized - they are given to movies by a set of experts HetRec2011 :: Said, Kille, De Luca, Albayrak Personalizing Tags 8 / 18
  • 12. Outline Movie Recommendation Folksonomies Our Approach Experiments Conclusion & Future WorkNot quite a folksonomy We have a problem: Tags are not personalized - they are given to movies by a set of experts We solve it: Tags are assigned ratings HetRec2011 :: Said, Kille, De Luca, Albayrak Personalizing Tags 8 / 18
  • 13. Outline Movie Recommendation Folksonomies Our Approach Experiments Conclusion & Future WorkPersonalizing Tags For each user, calculate the average rating for each tag based on the rating given to movies with each tag. HetRec2011 :: Said, Kille, De Luca, Albayrak Personalizing Tags 9 / 18
  • 14. Outline Movie Recommendation Folksonomies Our Approach Experiments Conclusion & Future WorkPersonalizing Tags For each user, calculate the average rating for each tag based on the rating given to movies with each tag. Little added effort if made at the time of the rating. HetRec2011 :: Said, Kille, De Luca, Albayrak Personalizing Tags 9 / 18
  • 15. Outline Movie Recommendation Folksonomies Our Approach Experiments Conclusion & Future WorkUsing tag ratings Append tag ratings to the user-movie matrix: HetRec2011 :: Said, Kille, De Luca, Albayrak Personalizing Tags 10 / 18
  • 16. Outline Movie Recommendation Folksonomies Our Approach Experiments Conclusion & Future WorkDataset www.moviepilot.de tag category # of elements % rating coverage 840 users Emotion 16 61.85 15, 613 movies Intended Audience 12 35.50 Place 763 75.39 33, 061 movie ratings Plot 5,565 90.00 6, 580 tags Time 224 64.02 HetRec2011 :: Said, Kille, De Luca, Albayrak Personalizing Tags 11 / 18
  • 17. Outline Movie Recommendation Folksonomies Our Approach Experiments Conclusion & Future WorkSparsity 100% 98% 97,46% 97,37% 97,40% 97,11% 96,50% 96% 94% Sparsity 92% 90% 88,59% 87,89% 88% 86% 84% 82% ratings +emotion +audience +place +plot +time +all HetRec2011 :: Said, Kille, De Luca, Albayrak Personalizing Tags 12 / 18
  • 18. Outline Movie Recommendation Folksonomies Our Approach Experiments Conclusion & Future WorkRecommender Collaborative Filtering kNN 50-fold random cross validation HetRec2011 :: Said, Kille, De Luca, Albayrak Personalizing Tags 13 / 18
  • 19. Outline Movie Recommendation Folksonomies Our Approach Experiments Conclusion & Future WorkResults 9,0E-5 2850% 8,0E-5 2500% 7,0E-5 Mean Average Precision 6,0E-5 5,0E-5 4,0E-5 3,0E-5 2,0E-5 1,0E-5 207% 296% 100% 162% 153% 0,0E+0 baseline emotion audience place plot time all HetRec2011 :: Said, Kille, De Luca, Albayrak Personalizing Tags 14 / 18
  • 20. Outline Movie Recommendation Folksonomies Our Approach Experiments Conclusion & Future WorkConclusion & Future Work Conclusion Simple additions to traditional algorithms generate large improvements Future Work Combinations of tags and time Tag-based recommendations for cold start users HetRec2011 :: Said, Kille, De Luca, Albayrak Personalizing Tags 15 / 18
  • 21. Outline Movie Recommendation Folksonomies Our Approach Experiments Conclusion & Future WorkThank you! Questions? HetRec2011 :: Said, Kille, De Luca, Albayrak Personalizing Tags 16 / 18
  • 22. Outline Movie Recommendation Folksonomies Our Approach Experiments Conclusion & Future WorkCaRR2012 2nd Workshop on Context-awareness 2nd Workshop on Context-awareness in Retrieval and Recommendation in in Retrieval and Recommendation in Conjunction with IUI 2012, Lisbon, Portugal conjunction IUI 2012. Content and Goals of CaRR 2012 Context-aware information is widely available in various ways and is be- Submission deadline: Dec. 2011 coming more and more important for enhancing retrieval performance and recommendation results. The current main issue to cope with is not only recommending or retrieving the most relevant items and content, but defining them ad hoc. Further relevant issues are personalizing and When: February 14th, 2012 adapting the information and the way it is displayed to the user’s cur- rent situation and interests. Ubiquitous computing furher provides new means for capturing user feedback on items and providing information. Where: Lisbon, Portugal The aim of the 2nd Workshop on Context-awareness in Retrieval and Recommendation is to invite the community to discuss new creative ways to handle context-awareness. Furthermore, the workshop aims on exchanging new ideas between different communities involved in URL: www.carr-workshop.org research, such as HCI, machine learning, information retrieval and rec- ommendation. Twitter: @CaRRws Important Dates (tentative) n Submission: End of Dec 2012 Program Committe (tentative) Omar Alonso • Linas Baltrunas • Li n Notification: tbd Chen • Brijnesh-Johannes Jain • n Camera Ready: tbd Dietmar Jannach • Alexandros n Workshop: February 14, 2012 Karatzoglou • Carsten Kessler • Antonio Krüger • Michael Kruppa Further Information • Ulf Leser • Pasquale Lops • Till nWeb: http://carr-workshop.org Plumbaum • Francesco Ricci • nE-Mail: info@carr-workshop.org Markus Schedl (to be extended) nTwitter: @CaRRws Chairs n Ernesto de Luca, TU Berlin n Matthias Böhmer, DFKI n Alan Said, TU Berlin n Ed Chi, Google HetRec2011 :: Said, Kille, De Luca, Albayrak Personalizing Tags 17 / 18
  • 23. Outline Movie Recommendation Folksonomies Our Approach Experiments Conclusion & Future WorkRecSysWiki www.recsyswiki.com HetRec2011 :: Said, Kille, De Luca, Albayrak Personalizing Tags 18 / 18

×