[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: ::scr googley eyed
On Tue, Oct 15, 2002 at 04:58:37PM +0100, Ben Evans said:
> As I may have mentioned / ranted about, collaborative filtering only works in
> places where you have enough of a repeating 'market'. It seems to me that
> certain types of searching are indeed this, but it's far from clear that
> *all* types of searching are.
True. Most of the work I've been doing has been on recommender systems
(like the aforementioned Amazon system) which do work best in a known
problem space.
However, for searching, I think that there are more creative ways to
improive the system than counting the number of links to a page and I
was mildly disapointed that nothing more creative came up in the google
programming competetion.
For example by doing various things to work out how similar a page is,
as discussed here ...
http://london.pm.org/pipermail/london.pm/Week-of-Mon-20021007/014117.html
(see the usual suspects turn up again and again :)
you could do a ranking on how popular somne information is - the same
Unix man pages, Linux how-tos and that joke about geek history
(http://www.dave.org.uk/geekhist.shtml for example) would keep popping
up.
Geographically clustering pages (the example that won the Google
Programming Contest is a cheap, unreliable trick but quite nifty) to
provide some localised information would be cool.
I have other ideas but I'm too tired to scribble them in this margin.
/me kicks the metaphorical mailing-list anthill to see if he can get
some other people to post.
Simon