Re: RSS feed filtered by keywords?

["jdklub <pseudonym@mindspring.com>" <pseudonym@mindspring.com>]

filtering RSS feeds (or any type of feed) is certainly useful. for 
example, a site that i may have no interest in reading on a regular 
basis may once or twice hit on a topic that i am interested in. i 
don't see any other way to find those needles in the big haystack, 
other then to sift and sort through all that data on a fairly regular 
basis. 

the first problem is creating the haystack. if you want as wide a 
reach as possible, i just don't see how individual desktop 
aggragtors/readers are going to manage. for one it takes time to 
fetch and index a couple thousand feeds. two, i'm sure RSS providers 
would frown upon everyone hitting every feed all the time, just to 
find the occasional item of interest. social networks may help solve 
this problem by creating a semi-focused haystack.

the other problem is the filtering. the /easiest/ way is keywords. 
however, as bill has pointed out, it's not the /best/ way to do it. i 
would say it's a good place to start. most people understand the 
concept of a keyword search. if they want to know about soccer, they 
type in "soccer" and the burden is going to be on the tools to be 
smart. users are not going to create fancy searches. 

this is why metadata is important, because it makes it possible to do 
clever things behind the scenes. something as simple as knowing what 
language a given item is in can be extremely useful. i've done 
keyword searches against a pool of several thousand feeds to look 
for "rss" and occasionally get results from other languages that (1) 
i can't read and (2) actually have nothing to do with RSS. if a 
particular result has information about it's trackback id, i can go 
find other items that may have not been returned in the original 
result, but might be relevant. something as simple as including the 
<dc:creator> element allows me to look for things written by people i 
am interested in. until people start including metadata (or i should 
say: until /tools/ start including more metadata) search tools can 
only get so creative. unlike google, an RSS search tool is not 
looking for /authoritative/ results, it's looking for /new/ results 
(that are still relevant). 

just some thoughts.