RE: [syndication] Re: XML-RPC and the Need to Cash In

["Gleb Dolgich" <glebd@kaunas.omnitel.net>]

I've found one solution to this problem. Novobot keeps local cache of
displayed headlines, and each channel has its own expiration day count,
after which its headlines are deleted from the cache (usually it is a
week or two). When a headline is displayed, it is added to the cache,
and if the same headline appears again, it is not displayed any more. I
don't know if this is a good approach, but it works quite well.

Gleb Dolgich
Proggle
http://www.proggle.com
Installer/GD - a better Windows installer
Novobot - the Web is news!


> I don't understand - in a browser, you can specify the display style
of links that have been visited. 
> If   the article has been viewed, the link looks different - you know,
blue underscore .vs. purple underscore.  
> Is this what you are talking about? or are you talking about how to do
this if you write your own UI?  
> like accessing (history) metadata about links in order to render them
yourself? 

That is one part of the problem I'm sure. However, as I understand it
the real problem is eliminating "redundant" information. In our
aggregator, I noticed the exact same links with the same headlines
popping up in different categories. Multiply that by perhaps a factor of
ten when you're dealing with weblogs, and you wind up with a lot of
redundant information.

One example may be Tomalak's Realm. You read a headline from TR and then
in the same application read 10 or 15 other weblogs. Each of those
weblogs also reads TR, and found that same headline interesting, so they
publish links to it as well. So now you have the exact same headline
repeated 10 or 15 times, and that's just one item!