“We think you’d also like…” and the Math of Suggestion - Part 2
gruenderszene.de
gruenderszene.de
http://news.ycombinator.com/item?id=548584
http://www.gruenderszene.de/it/we-think-youd-also-like-and-t...
As noted in the last round, this is a pretty light introduction to recommendations targeted mostly at only moderately technical folks (the audience of the blog that published it).
Anyone know a good way to get started - especially simple test data?
Toward the Next Generation of Recommender Systems: A Survey of the State-of-the-Art and Possible Extensions
http://portal.acm.org/citation.cfm?id=1070751
That one's fairly readable for people outside of the field. The Google News paper, which has some insights on doing large scale recommendations on a fairly dense user to item matrix, is a little more jumping into the deep end, but is worth glancing at even just to follow the references it sites:
Google News Personalization: Scalable Online Collaborative Filtering
http://www2007.org/papers/paper570.pdf
(The paper itself doesn't mention being problematic on sparse rating sets, but I've implemented something very similar and found that to be the case.)