Somewhat related, see this recent article (6/10/09):

"Extracting Meaning from Millions of Pages: University of Washington software pulls facts from 500 million Web pages."

http://beta.technologyreview.com/computing/22773/