Still skeptical about Music Recommendation Engines

I got a tweet from Tech Crunch’s Michael Arrington proffering An Automated Music Recommendation Engine that Actually Works.” This tweet linked me to this post on TechCrunch where it is reported that Muffin has an impressive pedigree and has a 3.5 million song database, and, again, that it works.

Well, given that this is a subject that I’ve pretty vocally voiced my skepticism about, both here on 9GS and on the TuneCore blog, I was intrigued.

Off to Muffin I went. For whatever reason, I tried to test it out by putting the band The Smiths into its search engine. I of course assume I’d get some recommendations for other mopey 80 rock bands. Imagine my surprise when instead of getting The Cure or Cocteau Twins or whatever, I got….J. Geils and the Smithereens. I can kind of see why it picked the Smithereens. Ya know: Smiths/Smithereens – they both begin with “Smith.” But we don’t need a complicated algorithm to do that, we basically need…um…a dictionary.

Anyway, more diligence is probably required, but I’d say this is fairly inauspicious.

Tags: , ,

  1. Bruce Warila’s avatar

    Nobody is going find recommendation engines very useful when they are attached to user interfaces like the one above. Watch the video attached to this link for another perspective.

    http://thesis.flyingpudding.com/?p=66

    Reply

Reply

Your email address will not be published. Required fields are marked *