Downloaded 12 times
![Evaluating playlist algorithms [M. & Lanckriet, 2011]
...
2. Compute playlist
1. Observe playlists from users likelihoods
?
>
3. Compare algorithms
by likelihood scores](https://image.slidesharecdn.com/hypergraphtalk-121015070828-phpapp02/75/Hypergraph-Models-of-Playlist-Dialects-3-2048.jpg)
![Evaluating playlist algorithms [M. & Lanckriet, 2011]
Key idea:
Playlist algorithm
=
Probability distribution
over song sequences](https://image.slidesharecdn.com/hypergraphtalk-121015070828-phpapp02/75/Hypergraph-Models-of-Playlist-Dialects-4-2048.jpg)
![Data collection
http://www.artofthemix.org/
Started in 1998, users upload and share playlists
[Ellis, Whitman, Berenzweig, and Lawrence, ISMIR 2002]](https://image.slidesharecdn.com/hypergraphtalk-121015070828-phpapp02/75/Hypergraph-Models-of-Playlist-Dialects-7-2048.jpg)
Uploaded byBrian McFee
Hypergraph Models of Playlist Dialects
The document discusses a study on playlist generation algorithms using hypergraph models, focusing on the evaluation of playlist diversity and the effectiveness of various features for generating playlists. It highlights experiments comparing global and categorical models, revealing that category-specific models and continuity are significant in achieving better accuracy. The dataset used, consisting of over 87,000 playlists and various song attributes, facilitates the exploration of these models' performance.
Hypergraph Models of Playlist Dialects
- 1. Hypergraph models of playlistdialects Brian McFee Lab Center for Jazz Studies/LabROSA Columbia University ROSA Laboratory for the Recognition and Organization of Speech and Audio Gert Lanckriet Electrical & Computer Engineering University of California, San Diego
- 2.
- 3. Evaluating playlist algorithms [M. & Lanckriet, 2011] ... 2. Compute playlist 1. Observe playlists from users likelihoods ? > 3. Compare algorithms by likelihood scores
- 4. Evaluating playlist algorithms [M. & Lanckriet, 2011] Key idea: Playlist algorithm = Probability distribution over song sequences
- 5. Modeling playlist diversity Playlists
- 6. Modeling playlist diversity Road trip Mixed Genre Party mix Hip-hop
- 7. Data collection http://www.artofthemix.org/ Started in 1998, users upload and share playlists [Ellis, Whitman, Berenzweig, and Lawrence, ISMIR 2002]
- 8. The data: AotM-2011 •98K songs indexed to Million Song Dataset • 87K playlists (1998-2011), ~210K contiguous segments • 40 playlist categories, user meta-data available
- 9. # Playlists percategory Mixed genre Theme Rock-pop Alternating DJ Indie Single artist Romantic Road trip Depression Punk Break-up Narrative Hip-hop Sleep Dance-house Electronic Rhythm & blues Country Cover Hardcore Rock Jazz Folk Ambient Blues 100 1000 104 105
- 10. # Playlists percategory Mixed genre Theme Rock-pop Alternating DJ Indie Single artist Romantic Road trip Depression Punk Break-up Narrative Hip-hop Sleep Dance-house Electronic • Majority of playlists are Mixed genre Rhythm & blues Country Cover Hardcore • Remaining categories: Rock Jazz Folk contextual/mood, genre, other Ambient Blues 100 1000 104 105
- 11. Our goals • Whichcategories can we model? Are some harder than others? • Which features are useful for playlist generation? • Do transitions matter? Are some categories less diverse?
- 12. A simple playlistmodel 1. Start with a set of songs
- 13. A simple playlistmodel 2. Select a subset (e.g., jazz songs)
- 14. A simple playlistmodel 3. Select a song
- 15. A simple playlistmodel 4. Find subsets containing the current song
- 16. A simple playlistmodel 4. Select a new subset
- 17. A simple playlistmodel 5. Select a new song
- 18. A simple playlistmodel 6. Repeat...
- 19. A simple playlistmodel 6. Repeat...
- 20. Connecting the dots... •Random walk on a hypergraph - Vertices = songs - Edges = subsets
- 21. Connecting the dots... •Random walk on a hypergraph - Vertices = songs - Edges = subsets • Learning: optimize edge weights from example playlists
- 22. Connecting the dots... •Random walk on a hypergraph - Vertices = songs - Edges = subsets • Learning: optimize edge weights from example playlists • Sampling is efficient, edge labels provide transparency
- 23. The hypergraph randomwalk model exp. prior edge weights transitions playlists
- 24. Edge construction: example •Audio: cluster songs by timbre
- 25. Edge construction: example •Audio: cluster songs by timbre Audio-1 Audio-2 Audio-4 Audio-3 • Multiple clusterings (k=16, 64, 256)
- 26. Edge construction: thekitchen sink • Audio • MSD taste profile • Era • Familiarity • Lyrics • Social tags • Uniform shuffle • Conjunctions: "TAG_jazz-&-YEAR_1959" • 6390 edges, 98K vertices (songs)
- 27. Evaluation protocol • Repeatx10: - Split playlist collection into 75% train/25% test - Learn edge weights on training playlists - Evaluate average likelihood of test playlists • Compare gain in likelihood over uniform shuffle baseline
- 28. Experiment 1: globalvs. categorical • Fit one model per category • Fit one global model to all categories • Test on each category and compare likelihoods • Question: When does categorical training improve accuracy?
- 29. Experiment 1: globalvs. categorical Unifo rm ALL Mixed Global model Theme Category-specific Rock-pop Alternating DJ Indie Single artist Romantic Road trip Punk Depression Break up Narrative Hip-hop Sleep Electronic Dance-house R&B Country Cover songs Hardcore Rock Jazz Folk Reggae Blues 0% 5% 10% 1 5% 20% 25% Log-likelihood gain over uniform shuffle
- 30. Experiment 1: globalvs. categorical Unifo • Largest gains for genre playlists rm ALL Mixed • No change for "hard" categories Global model Theme Category-specific Rock-pop Alternating DJ (e.g., Mixed, Alternating DJ, Theme) Indie Single artist Romantic Road trip Punk Depression Break up Narrative Hip-hop Sleep Electronic Dance-house R&B Country Cover songs Hardcore Rock Jazz Folk Reggae Blues 0% 5% 10% 1 5% 20% 25% Log-likelihood gain over uniform shuffle
- 31. Experiment 1: learnededge weights ALL Mixed Theme Rock-pop Alternating DJ Indie Single Artist Romantic RoadTrip Punk Depression Break Up Narrative Hip-hop Sleep Electronic music Dance-house Rhythm and Blues Country Cover Hardcore Rock Jazz Folk Reggae Blues Audio CF Era Familiarity Lyrics Tags Uniform
- 32. Experiment 2: continuity? •Do we need to model playlist continuity? edge weights songs • Simplified model: - ignore transitions - choose each edge IID exp. prior playlists • Question: Are some categories more diverse than others?
- 33. Experiment 2: continuity Unifo rm ALL Mixed Global model Theme Category-specific Rock-pop Alternating DJ Indie Single artist Romantic Road trip Punk Depression Break up Narrative Hip-hop Sleep Electronic Dance-house R&B Country Cover songs Hardcore Rock Jazz Folk Reggae Blues -15% -10% -5% 0% 5% 10% 15% 20% Log-likelihood gain over uniform shuffle
- 34. Experiment 2: continuity Unifo rm ALL Mixed Global model Theme Category-specific • Most categories exhibit both Rock-pop Alternating DJ Indie continuity AND diversity Single artist Romantic • Transitions are important! Road trip Punk Depression Break up Narrative Hip-hop Sleep Electronic Dance-house R&B Country Cover songs Hardcore Rock Jazz Folk Reggae Blues -15% -10% -5% 0% 5% 10% 15% 20% Log-likelihood gain over uniform shuffle
- 35. Example playlists Rhythm & Blues EDGE SONG 70s & soul Lyn Collins - Think Audio #14 & funk Isaac Hayes - No Name Bar DECADE 1965 & soul Michael Jackson - My Girl Electronic music EDGE SONG Audio #11 & downtempo Everything but the Girl - Blame DECADE 1990 & trip-hop Massive Attack - Spying Glass Audio #11 & electronica Björk - Hunter
- 36. Conclusions • Category-specific modelsoutperform global playlist models. • Continuity matters! • Proposed model is simple, efficient, and transparent • AotM-2011 dataset available now! http://cosmal.ucsd.edu/cal/projects/aotm2011
- 37.