Should We Care About Content? Recommending by Proxy with Big Metadata

When constructing a music recommender system, which is more important: a musicological understanding of the catalog of music in a system or the number of times two particular songs were played one after the other and were `liked’? Even better, if a system knows the latter, does the former even matter? Do machines that predict behavior need to learn to listen? Or is observing behavior enough?

Transcript

1. Should we care about content? Recommending by proxy with big metadata B e n F i e l d s @ a l s o t h i n g s
2. Goal:
3. how to make a recommender
4. not so much how
5. Rather: ‘what’
6. what data should be used to make a recommender?
7. First, Examples
8. Image
9. http://www.flickr.com/photos/dexxus/5652914929/
10. http://www.flickr.com/photos/msojka/5285298402/
11. http://www.flickr.com/photos/pinkertons/7730547912/
12. http://www.flickr.com/photos/nicholas_t/626009491/
13. http://www.flickr.com/photos/dexxus/5652914929/
14. http://www.flickr.com/photos/dexxus/5652914929/ http://www.flickr.com/photos/msojka/5285298402/ http://www.flickr.com/photos/pinkertons/7730547912/ http://www.flickr.com/photos/nicholas_t/626009491/
15. http://www.flickr.com/photos/dexxus/5652914929/ http://www.flickr.com/photos/msojka/5285298402/ http://www.flickr.com/photos/pinkertons/7730547912/ http://www.flickr.com/photos/nicholas_t/626009491/ Same Camera Model Colour Palette Distance Location
16. Music
17. A small playlist
18. What should follow this? Meat Loaf - I'd Do Anything for Love
19. maybe? Beethoveen’s Piano Sonata No. 14 in C Sharp Minor, Op. 27, No. 2
20. how about? Journey - Don’t Stop Believin’
21. A small playlist Meat Loaf - I'd Do Anything for Love Beethoveen Journey
22. Recommending?
23. prediction of opinions
24. prediction of opinions about things
25. prediction of Actions on things
26. Metadata?
27. Data About Data
28. Content descriptors are metadata
29. Ratings are metadata
30. But it isn’t always neat and Tidy
31. human curated Content descriptors are metadata
32. In the wild
33. @alsothings 33
34. @alsothings 34
35. @alsothings 35
36. @alsothings 36
37. @alsothings 37
38. @alsothings 38
40. Contests
41. Netflix Prize
42. Netflix Prize 500K people’s ratings of 18k movies Take these and predict the ratings of movies that haven’t been Rated
43. BellKor
44. gradient boosted decision trees
45. Any Value in Content?
46. No.
47. Million Song Dataset (kaggle)
48. Million Song Dataset challenge (kaggle) The partial listening history of 1M people Predict the tracks that are missing
49. Collaborative filtering
50. Any Value in Content?
51. No.
52. so?
53. moar data > deeper Analysis
54. Simple use of metadata from far and wide gets you further than Deep understanding of core-data
55. But it doesn’t help you explain things to the user.
56. http://www.netflixprize.com/community/viewtopic.php?id=1537 http://www.kaggle.com/c/msdchallenge http://mir-in-action.blogspot.co.uk/2012/09/collaborative- filtering-still-rules.html http://www.dtic.upf.edu/~ocelma/PhD/ http://www.slideshare.net/plamere/music-recommendation- and-discovery Ben Fields @alsothings Questions?