I've been doing some work recently on predictive newsreading ... adapting open source software to give people a customized daily newspaper of online news tailored to their interests. As a big fan of citizen journalism I've been focusing on blogs as news sources, where RSS feeds are ubiquitous.
A colleague showed me the AideRSS release today, which looks like a great start in prediction!
At first blush AideRSS appears to accept a list of RSS feeds, which can be blogs or aggregate or customized newsfeeds like canned Google Searches, apply a "black box" ranking algorithm to the feeds, and return a subscriber a customized feed with ranking attached.
The subscriber can then apply filter criteria to the custom feed on a per source basis to get a filtered custom feed that is, hopefully, predictive.
There are obvious next steps, the first of which is probably related to opening and customizing the ranking algorithms (spare change for development, anyone?) but the concept is pretty interesting, especially as it is in production, and it's easy enough to test out.
Try this for a good time:
Now ... paste the URL of your aggregate filtered feed back into your reader as a new feed.
You can now compare the list of filtered and unfiltered news in a single interface, tuning the filters to suit your likes