MusicBrainz Rating System
This page is not complete yet. WikiWarden
s: please ignore this page for the next day or so.
As part of the Google Summer of Code program, Sharon Myrtle Paradesi is creating a collaborative filtering system for MusicBrainz. This collaborative filtering system will allow us to replace the old and decrepit artist to artist relationships system we currently have. The new system will allow the MusicBrainz community to create similarities not only between artists, but between labels, releases and track. Knowing how the various pieces of music inside of MusicBrainz relate enables us to:
- Partition the database into more managable chunks. Right now peer reviewers (voters) will be presented with all database changes, even for languages/scripts that they may not know about. With stronger data relationships we can present more tailored edits to our peer reviewers and lessen the burden of peer review.
- Provide music discovery services. Collaborative filtering allows us to create features that lets users discover music similar to Amazon's recommendations: If you liked, this then you'll like that. However, unlike Amazon's system our system will be without a bias to sell crap.
- Make the MusicBrainz data more attractive to commercial uses. Commercial uses, like the BBC using the MusicBrainz data, brings in money to the MetaBrainz Foundation, which will in turn pay for more developers, more servers and hopefully some ponies.
The first step towards building a collaborative filtering system is a comprehensive rating system that allows users to rate artists, labels, releases and tracks based on their personal taste. This new rating system (along with folksonomy tagging) is essentially the first foray into subjective metdata for MusicBrainz and is therefore going to be considered experimental. At this point we do not know if this is going to be a good fit for MusicBrainz or if it is going to work at all. However, using the same rationale that we used for creating MusicBrainz in the first place, we're going to try it and if it doesn't work you'll tell us. And if the answer is that it doesn't work at all, then we'll toss it out.
It may also turn out to be the wrong set of features for MusicBrainz. Some people have suggested that these features might best be suited for a new site altogether that does not focus on editing data, but solely on rating/tagging music. We'll have to see how these new features pan out before we can really decide the proper course of action.
In the very first round of feedback on this feature we've gotten a wide range of feedback ranging from "I hate it, its the wrong thing for MusicBrainz!" and "Wow -- that's cool!". Given that, we're going to add an option for turning off and hiding ratings altogether so that you will not have to look at ratings if you don't want to. I would very much like to ask people to give us the benefit of the doubt on this new project. We would, however, love to take any constructive feedback/criticism you have for us.