Difference between revisions of "2009:Music Recommendation"
From MIREX Wiki
(→Potential Participants) |
(→Potential Participants) |
||
| Line 29: | Line 29: | ||
# Koosha G, University of Alberta | # Koosha G, University of Alberta | ||
| − | # Nicolas Wack et al., MTG Universitat Pompeu Fabra, Spain, nicolas[dot]wack[at]upf[dot]edu (2 algos presented) | + | # Nicolas Wack et al., MTG Universitat Pompeu Fabra, Spain, nicolas[dot]wack[at]upf[dot]edu (2 algos presented, 1 by me and 1 by Dmitry Bogdanov) |
Revision as of 10:05, 6 August 2009
Contents
2009 Music Recommendation Task Overview
This task is for evaluating tasks that produce a list of songs from an input. This is a new task for 2009.
Task Overview
Given a query and a set of source data, produce an ordered list of songs. The results are evaluated against a ground truth derived from a second source or human evaluation.
Potential subtask Types
Personalized Radio Song Similarity
Potential Source Data
- Audio Content - Which data set of audio files should be used?
- Artist Tag Info - set of all tags associated with the artist that performed the audio (LastFM webservices)
- Song Tag Info - set of all tags associated with this song (LastFM webservices)
- User Profile - Profile information of a LastFM user that has listened to at least a subset of the target audio (LastFM webservices)
- User playcounts - Play counts of audio content for a particular user (LastFM webservices)
- Playlists - online sources of playlists containing the targeted audio (location???)
Evaluation
Potential Participants
- Koosha G, University of Alberta
- Nicolas Wack et al., MTG Universitat Pompeu Fabra, Spain, nicolas[dot]wack[at]upf[dot]edu (2 algos presented, 1 by me and 1 by Dmitry Bogdanov)
