millionsong's blog

Hi everyone, it's been a while since we posted news regarding the MSD, so here are thoughts on two recent items: Serra et. al's work in Scientific Reports, and the end of the Million Song Dataset Challenge.

After a few weeks of competition, top contestants on the Million Song Dataset Challenge seem to have reached a plateau around 0.15 mean average precision (MAP). It is impossible to say at this point what method they use to achieve that score, but there is a good chance that this represent the best score obtainable through collaborative filtering (CF). Now what?

Short post, a formal announcement will be sent on music-ir and other mailing lists, but the contest is live! Check it out on Kaggle.

Next week I'll be representing Brian McFee at AdMIRe 2012 in beautiful Lyon.

Wondering what does a network of a million songs look like? Well, visrel.com has the answer.

Yesterday I attended CIRMMT's large amount of data workshop where I ranted about song matching. Here are my slides, you can see a naked man and a cute dog, among other things.
--TBM

Following up on our previous post, we worked to resolve some unexpected matches in the Taste Profile subset. However, the following result is important for anyone working with the song or artist metadata in the MSD.

As we all agree, every new band or artist should get assigned a MD5 code and use it for the rest of their careers. Titles should be limited to ASCII characters, less than 50, no weird punctuation or parenthesis. Any violation shall be punished by paper cuts.

If you happen to be near San Diego around February 9th (Thursday), you probably want to see this MIR All-Star roster at ITA, organized by UCSD: http://ita.ucsd.edu/workshop/12/talks

Don't you know 'Harmonia'? According to numbers compiled by Paul Lamere from the recently released Taste Profile Subet, you should, they're super popular!