An excellent and comprehensive article on filtering spam using Bayesian Statistics by Paul Graham. [Found via 0xdecafbad].

I am pleased to annouce the release of Aggie 1.0 RC4.

Special thanks to: Simon Fell, Eric Vitiello, Ziv Caspi, Ingve Voremstrand, and everybody that has submitted bug reports and enhancement requests.

Highlights for the RC4 release

Internationalization

Aggie now has better support for international feeds.

Proxy Support

Now you can use Aggie with a proxy server.

Stylesheets

Aggie supports having different stylesheets to style the resulting HTML. Each stylesheet can have it's own configuration dialog and saved parameters. Here is a detailed explaination of how this system works. How Templates Work In Aggie.

Speed Improvements

There have been many improvements that help Aggie get the news faster. There is now support for the "If-Modified-Since" HTTP header, meaning Aggie will only download a file if it has changed since the last time it was retrieved. Also the HTTP Redirection header is obeyed and stored to reduce the number of round trips retrieving files. In addition all this information is stored in a seperate file so that if you are using Aggie to read an OPML file from a remote site you still get all the speed enhancements.

Command-Line

Aggie now ships with a command line version of Aggie that can be used in background tasks.

Referer Logs

Aggie now supports placing a link back to the users website in the referer logs. The genesis and explaination of this idea is documented on Content Syndication with XML and RSS. The page that the referer log entry points to is here (using bitworking.org as the example site) http://bitworking.org/AggieReferrers.html?userWeblog=http://bitworking.org

ThinRSS (and others doing xml UI definitions in Java) inspired me to having a look at how easy this sort of thing is in Zeepe.

The result is here: Zeepe Applications.

Zeepe currently doesn't support XML UI definitions - because XHTML is already an XML UI definition.

A list of rules to follow to auto-discover rss feeds: Ultra-liberal RSS locator

---

[Scripting News] : Related to that, where are the developers these days. I'm not talking about hoardes of people who clone Unix and Unix utilities. I mean people doing real new software, new ideas, patentable stuff, who aren't taking the patents. Those are the people we should be hearing from. I also like hearing from smart respected lawyers. I'm not one of those people who think all lawyers are slime. But something is really wrong when all we hear from re technology are lawyers. That's when you get disconnects like his oft-repeated mantra that developers aren't doing anything. Well, Larry, if you don't talk to developers, how could you possibly know?

---

All too busy to bother e-mailing? There's loads of software out there, what does he want to hear?

Flash Explores New Angles. Macromedia's Flash MX allows users to do a number of cool tricks, including switching instantaneously between video clips. By Lisa Delgado. [Wired News]

What I'd like to see is this done in a really cute Zeepe wrapper frame.

Via Radio Free Blogistan to organica who kindly provide a list of weblog crawlers - what do they do beyond crawl and simple search?

Jon Udell's (Personal RSS Aggregators) article discusses the benefits of what I call "citation analysis" - a number of people are referring to this, but all the context of Google. While probably the most famous implementation I think it is quite probable that weblogs themselves will benefit from this analysis to determine "hubs" and "authorities".

Profundis (it no longer works, but the docs are 'interesting') implemented a variation on the algorithm proposed by Jon Klienberg and yielded interesting results. What Profundis showed was the web as communities - and the algorithm would reach a steady state on these communities. What would be interesting is whether weblogs are also  "steady state" communities (they should be because they are just web pages) and whether application of the algorithm would enable one to find (by recommendation of the algorithm) the most relevant to oneself of the logs in the community.

Profundis was written a couple of years ago - it appears that Jon Klienberg has some new papers up (or after I was looking) on gossip protocols (whatever they are) and analysis of small worlds - might be interesting to follow up.

The trouble is that once you start wandering around the web it consumes nearly all your time (which is OK for journalists) and you get sunk beneath 8 million different possible ways of implementing "hello world":

http://www.mozilla.org/projects/xul/xre.html

This is and will be an interesting thing to keep an eye on: Mono.