I never heard about GroupLens. From what I red on Wikipedia, Usenet is the only community discussion system to which collaborative filtering was applied. GroupLens is interesting, but I'm afraid porting it (if that makes any sense) to an Internet forum or mailing list would hardly save efforts on implementing something based on a CF library. That said, I don't think CF is at all necessary for a good message rating system, though data on the efficiency of CF in discussion systems would be interesting. I also don't think the idea proposed is opposed to CF; the existence of a median ("common") value does not prevent the generation of personalized ratings. For now I see CF as a cherry on the top.Filipus Klutiero <chealer@gmail.com> wrote: [...]> I'm not aware of any software with such a feature that would fit for > Debian. I also couldn't find any in a quick search. [...]It sounded a lot like the old GroupLens usenet tool to me. http://www.gnu.org/software/emacs/manual/html_node/gnus/GroupLens.html The big difference there was that it explicitly grouped your ratings with those of other people who rated it the same way - there was not the assumption of a common value system which seems to underlie this proposal.
Unless you're suggesting a recommendation system, I don't see how this paper is relevant.I see that http://www.grouplens.org has moved on and their research (like http://www.grouplens.org/node/126 - "The recommendations that are most accurate according to the standard metrics are sometimes not the recommendations that are most useful to users") may be informative for anyone who thinks that message-voting will surely work.