Paul Lamere is the principal investigator for an interesting project over at Sun Labs called "Search Inside the Music." It's stated goal is "to explore new methods of categorizing, indexing and organizing large collections of music to allow more effective ways of searching through these collections." I know that I have an ongoing background task to add metadata to my music as a way to create useful playlists semi-automagically, but it never seems to be enough. So anything along these lines that worked well would a real nice-to-have.
The project focuses in part on using acoustical similarity algorithms to help group together music. The idea is that if you like, say, Jefferson Airplane you may like other music with that psychedelic rock sound. What's just as interesting though is that Paul's project is now also looking at using social data to recommend and organize music based on the preferences of those with similar music tastes.
In fact, over on his blog, Paul makes the provocative statement:
Recommendation is the next 'search'. Companies are trying hard to give people good recommendations - the web is becoming populated with music recommenders such as last.fm, myStrands, Qloud, and iLike. Companies like netflix are offering million dollar prizes for improved recommendations. If you are building a site with lots of content - whether it is music, video, blogs or dinner recipes - you will do well to include a recommender.
I've been a bit cool to the whole tagging phenomenon especially when it comes to attaching metaphysical significance to, e.g., a tag of "film" vs. a tag of "movie." (Folksonomies was one term from the early tag hype that--fortunately--seems to have departed the stage after its 15 minutes.) However, as these various networks of music, photos, bookmarks, video, etc. grow, human recommendations and rankings are clearly emerging as at least one important way of making sense out of the stupefying quantity of data in the network.