COMPUTER scientists have long tried to foist order on the explosion of data that is the internet. One obvious way is to group information bytopic, but taggingit all comprehensively by hand is impossible. DavidBlei, of Princeton University, has therefore been trying to teachmachines to do the job.
