On Sun, 31 Dec 2000, Aaron Voisine wrote:
> The thing I really don't like is the idea of people picking their own
> keywords. That was tried with HTML meta tags. People who want
> their information viewed (the entire marketing industry) have a vested
> interest in puting in every single popular keyword whether it relates
> to their info or not. This problem is greatly reduced, although not
> eliminated, by searching on the name of the file.
Yeah. This is also solved by yahoo type indices, where information is
mediated by the yahoo gods. There have been vote counting proposals to
address the quality of information placed on Freenet. The Google
technique of raising a web pages score when a lot of links are referencing
it is another way of automatic vote counting.
> Also with user defined keywords, you have to concern yourself with
> plurals and other forms of root words. If a document is listed under
> the keyword "cars" and a user searches for "car", they won't find it.
Human keyword assignment is labor intensive and error prone, but Humans
are still very good at generalizing. With a general purpose information
network like Freenet, there should be many ways to index different types
of media. For example, an mp3 file is better indexed by cddb lookup and a
stock quote is better indexed by its ticker symbol. Packaging good
indexing tools with Freenet is important to it's viability as a
information resource. The MMK@ key type for map files and associated
tools for generating them is a prime example of this.
_______________________________________________
Freenet-dev mailing list
[EMAIL PROTECTED]
http://lists.sourceforge.net/mailman/listinfo/freenet-dev
