Newsblaster (Idle Words)#

04.30.2003

Columbia has a news digest site they call Newsblaster which is the best I've ever seen. Each news category has a summary auto-synthesized out of a slew of articles that their crawler finds. Unlike Google news, which just pastes together the lead sentences of several stories, Columbia does some sophisticated processing on the text to actually pick out and rank the most relevant sentences. Then it figures out how to stitch them together.

Be sure to check out the /dev/ directory on their site, where they do multi-lingual news summaries, too. Nice!

Apparently this stuff is free and open, we're working on trying to get our hands on it. You can imagine how useful good auto-generated summaries would be in a blog aggregator/categorization tool.