The MTBoS Genome Project
Have you ever listened to Pandora and wondered what method they used to determine what songs to play for you? I did and remember writing a research paper about it back in grad school.
Pandora makes use of something called the Music Genome Project. Professional musicians will actually listen to every song in their database and tag all of the songs along different dimensions – timbre of the instruments, vocal type, volume, bpm, etc. Each song then gets a vector associated with it where each dimension is one of those categories.
Then, when you put in a seed song, Pandora will calculate the closest songs to your seed, basically using the distance formula in hundreds of dimensions. (There’s some weighting and tweaking, of course, but that’s the core premise.)
—
At the Sunday My Favorites session of TMC14, Bob Lochel and Megan Schmidt show us how to find our closest buddies by filling out a survey about what movies they like. Then they calculated the correlation coefficient and the people who correlated the most were the best friends of the pair.
On Saturday, I was talking with someone (I think it was Matt Baker) about how to help people get into our community. He mentioned that while there are a lot of good ideas out there, the ideas that resonate the most with him are the ones that comes from they people he most identified with – whose teaching style was most like his. I had the idea that we could somehow make a survey that a new person could fill out and it would give them a personalized output of Twitter accounts and blogs to follow – a somewhat advanced version of the category lists we made here during TMC12.
—
I want to make this, but I need help. What are the dimensions that we should ask about? What are important aspects of your teacher identity, and what are some of the things that make you feel on the same wavelength as another teacher in the MTBoS? Please let me know so I can start compiling these dimensions and building this.
(Also, if there is anyone more skilled in programming who is willing to help me, ping me.)