Re: WEB: Yahoo/Firefly Website recommendation service
At 06:52 PM 12/11/96 -0800, Dale Thorn wrote:
Alexander Chislenko wrote:
Firefly Network Inc. has just launched a public beta of our website recommendation service on My Yahoo! This service is the result of a partnership between Yahoo! Inc. and Firefly Network, Inc. in application of Automated Collaborative Filtering (ACF) technology to the Web. It allows users to find interesting websites interest and like-minded people, and otherwise help the user navigate the vast domain of sites and people in an intelligent and personalized way.
I tried Firefly. What a waste. Unless your tastes in most things entertainment-wise are pretty mundane (music=Pearl Jam, Prince, other mainstream drek), they won't be able to find a match at all, no matter how much information you give them!
You are talking about our first site, with music and movie recommendations.
I personally rated not-at-all-mainstream movies by Kurosawa and Tarkovsky and
got a similar advice. Though, I agree, the recommendations are not perfect.
Partly, because the most of the audience are mainstream. Partly, because
the site
Alexander Chislenko wrote:
At 06:52 PM 12/11/96 -0800, Dale Thorn wrote:
Alexander Chislenko wrote:
Firefly Network Inc. has just launched a public beta of our website recommendation service on My Yahoo! This service is the result of a partnership between Yahoo! Inc. and Firefly Network, Inc. in application of Automated Collaborative Filtering (ACF) technology to the Web.[snip] The website recommendation uses the feature-guided ACF server that I hoped would solve most of the algorithmic problems we encountered in the music and movie domains. I'd be interested to know what you think of it.
Thanks for the reply. My experience with Firefly #1 has shown that the people who would most need the service (or need it at all) are the least likely to benefit from it, which is why I think it's a bad thing, wasting a lot of valuable time hand-entering data, only to find it doesn't correlate the more interesting (non-mainstream) data. I'll try the website service, but my up-front instinct is that it won't be much better unless the software is a *whole lot* better.
participants (2)
-
Alexander Chislenko
-
Dale Thorn