News@Nature / Postgenomic mashup
pgrelevant.user.js installs a command under "User Script Commands" in the Tools menu called "Find relevant blog content" which is available on pages from News@Nature. If you click on this command a bar will appear at the top of the current page and you'll see a list of key terms found followed by a list of relevant blog posts (that's the theory, anyway).

The script:
1) Extracts what it thinks is the main body of the page
2) Sends it to the Yahoo! term extraction API
3) Sends those terms to the Postgenomic API
4) Displays the resulting blog posts
All using the magic of JSON and Javascript.
The script's main limitation (and the reason that it's currently restricted somewhat arbitrarily to news@nature pages) is step 1: extracting content from a web page is, er, remarkably difficult. Noise from elements like scripts, navigation bars and adverts render the term extraction API useless. Perhaps somebody with better Javascript skills could help with that...
