If your literature research takes you back any earlier than, say, 1900 or so, you have probably had occasion to try to decipher old type faces, older spelling variants, and blurry, damaged or fading text.  Many of these kinds of texts are being digitized, however, optical character recognition (OCR) scanning has a hard time dealing with anything that is not clearly printed with current spelling or type face conventions.  Even with more current texts, machine digitization (like the Google Books project) introduces a high number of errors to the electronic “copy.”  Having a human go through and correct for these errors is extremely labor-intensive and therefore costly.  A group at Carnegie Mellon has figured out a way to harness human powers of deciphering in order to read and digitize old texts.  Read here for more information about reCAPTCHA.

Leave a Reply