The Reuters news API mash-ups.

Using Open Calais to extract terms from Reuters news articles, this breaks news down into people, places, organisations etc, and then aggregates stories at these levels, bringing the most popular stories to the top. Great prototype, lots of scope here...

"top down semantic solutions". this space is becoming very 'enterprise' and corporate very quickly. which is probably a good thing for solutions that work, but what are the algorithms hard coding into our information economy?