ollieglass

I'd love to do something like this in London. Inspired by the manual, I've created a signup page at hackerschoollondon.launchrock.com Perhaps we can get one going here if there's some interest.

n.b. Hacker School have asked me to stop using their name - I don't mean to mislead or imply any affiliation with them. I'll change it shortly. But I do want to copy the idea :)

You're not being dumb, it's currently limited to the top 10 results.

I've not. A Spotify would be a nice feature.

Would you use an API?

No, just MySQL.

I'm using the part of speech tagger from Python's NLTK to find the adjectives (http://nltk.org/). Sometimes it selects phrases instead of individual words, I'm not using any extra rules to do this.

I've left the phrases in because I think they're interesting and informative. Do you agree, or do you find them distracting?

Not sure about your side note!

I wrote Akin To to scratch a personal itch - I find music recommendation systems really boring. I search for John Coltrane, they all recommend Miles Davis. That's obvious and uninteresting.

I wanted to make a system that could suggest Squarepusher and Steve Reich by using the underlying themes and qualities of the music to make relevant and novel recommendations.

This thematic information already exists in music reviews, I've extracted adjectives from them and used them as a way to search, explore and compare album reviews. It kind-of works. I'd love to hear your thoughts and feedback.

Thanks!

Not yet, I'd love to try it with other sites and types of reviews.

Really? Could you give more detail - at what level of traffic do they shut down links?

Are there any technical details available of how the system functioned? And has anything similar been attempted since?

Could you say what differentiates it from Backbone?