I wrote here over a year ago about my love for collaborative filters, the use of informatics to collect information and provide recommendations based on that collection. At the time, I said that I was using Amazon and RateYourMusic for filling my music needs, and that still stands. In fact, I've been working on "normalizing" my ratings: I've defined a 10 point bell curve that matches how I keep track of my track-by-track ratings of all these songs in iTunes (using the id3 comment field, since iTunes's internal rating system uses only a 0-5 star rating scheme, which wasn't finely enough detailed for me). This is my rating scheme, with the rough descriptions of what each rating means:

I can export my iTunes data into an Access database, then do some grouping where I determine the collective ratings by artist and album, thus creating the data by which I keep my favorite albums page updated.

Brian Almquist mentioned the AudioScrobbler site to me when we met last year to discuss collaborative filters, but at the time it wasn't doing recommendations and I was still using MusicMatch Jukebox as iTunes hadn't been released for Windows yet. Both of those things have now changed, and so a few days ago I signed up for AudioScrobbler and have been submitting data, and am now anxiously awaiting the first group of recommendations for artists that I might like.