h a l f b a k e r y
Still more entertaining than cricket.
add, search, annotate, link, view, overview, recent, by name, random
news, help, about, links, report a problem
or get an account
First off I have to admit that this is (was) a baked idea as I have seen it before, but can not seem to find it now.
A web site where you are prompted to enter 3 or 5 or 10 CD's that you like. (Artist and Album name) You are returned a list of other CD's that you will probably like (sorted by likelyhood
The results are based off of the entries made by other people when requesting a list. (Your entry list is incorporated into the database as a group of CD's that are "alike" (they can be assumed to be similar in the sense that one person likes all of them.)
The Engine is sort of self purpetuating, as more people use the database, the data is further refined and expanded.
Would work best if you enter you lists in genre's (i.e. when looking for a mellow CD, enter you favorite mellow CD's)
I have only given an occasional thought as to the algorithym to use for this and how to store the inter-relationships. If anything I think this would be a interesting project to work on.
(Also could be used for Books and Movies) I realize that Amazon and IMDB have recomendations but I do not trust the integrity of Amazon (they think I will like whatever they have extra copies of) and IMDB recomendation is based off of one movie at a time.
Same idea, for movies [Uncle Nutsy, Aug 11 2000, last modified Oct 05 2004]
collaborative filtering music
Music recommendation is a canonical application of collaborative filtering. Not quite "baked" but at least attempted. [egnor, Aug 11 2000, last modified Oct 05 2004]
This site does a pretty decent job but the UI sucks! [dgeiser13, Aug 11 2000, last modified Oct 05 2004]
Collaborative Filtering Resources
Haven't had a chance to peruse all of these yet...one of them might do music!!! [dgeiser13, Aug 11 2000, last modified Oct 05 2004]
does exactly what you asked for [neilp, Dec 21 2004]
Audio Scrobbler, if you haven't you should
does what you should have asked for [neilp, Dec 21 2004]
||You probably remember Ringo a.k.a Firefly a.k.a HOMR, research that came out of the MIT and ended in a startup that was bought by Microsoft and never heard of again.
||I agree that having an independent, IMDB-like entity maintain this would be better. This really seems like something that should exist and could easily support itself from ad revenue, and I don't really understand why it doesn't. Patent issues on the similarity engine algorithm, maybe?
||Something very much like this exists for movies at http://www.moviecritic.com. It works very well and has all sorts of bells and whistles, like enabling two users to ask what movies would best suit them as a pair or letting you check for the *worst* things you could see.
||They're interested in licensing their similarity engine algorithm, too. This could work...
||Offtopic: Firefly was bought by Microsoft for their "Passport" universal site login nework service (i.e., login once centrally, never have to keep track of site-specific logins again, the central authority enforces privacy policies); it was considered at the time that recommendation or "collaborative filtering" services were a failed idea.
||(I'm not sure if that's really true, but it's certainly the case that you don't hear much about CF any more. There is some patent coverage, but that's not really the problem; it's just not very cool anymore. Microsoft themselves did a lot of research on them -- I even implemented a prototype recommendation service for Sidewalk when I worked there! -- before dropping the notion. As noted, Amazon does use them.)
||Personally, I think CF just plain doesn't work.
||True to their intentions, Microsoft is finally bringing out Passport again (though it took a bit of confused muddling first).
||In a final offtopic note, do you people realize that Amazon owns the IMDB? Look at the very bottom of the IMDB home page.
||"I'm not in this for the money. This is all about building jobs... "
-- Nick Grouf, Chair & CEO of PeoplePC, who launched his company with 50
employees and $65 million in venture capital. Grouf sold his previous
business, Firefly, to Microsoft for at least $30 million . . . all in an
attempt to "democratize technology." (San Francisco Chronicle, 2/5/00)