Monday, July 26, 2010

The complicated world of recommendations

Laura Miller at Salon discusses some options: 'Amazon and other online merchants have harnessed mighty algorithms to run their "If you enjoyed that, you might like this..." suggestion engines, but these are still crude instruments. Practically any novel you plug into Amazon's search engines at the moment returns the robotic announcement that people who bought it also bought one of Stieg Larsson's "Girl" thrillers — because seemingly everybody in America is buying those books. It's not like you need the world's most sophisticate e-commerce servers to tell you that. Recognizing that book recommendation may as yet defy science, a couple of literary types are currently offering artisanal advice.' More...

Meanwhile, the whole business of recommendations comes up a different way in this month's issue of Wired, which, unfortunately, isn't online yet. Then again, I read it in a real magazine and so could you! What a concept. Anyhow, the article discusses in length the engines of recommendation based on known data. It's a tough business, overall, and a fascinating one.

No comments: