ACM TechNews

ReCaptcha: Reusing Your 'Wasted' Time Online

CNet (07/16/08) Olsen, Stefanie

The goal of the ReCaptcha project is to use captcha technology--distorted word puzzles that humans can successfully solve but machines such as spam bots cannot--to improve machines' identification of scanned text that a computer has trouble recognizing optically due to faded ink or blurriness, so that print archives can be more effectively mined by search engines. Carnegie Mellon University (CMU) professor and ReCaptcha creator Luis von Ahn says up to 600 million people have completed at least one ReCaptcha on sites that use the technology in the last year, and such activity is aiding and expediting ambitious text-scanning initiatives such as the New York Times digitization project. Von Ahn debuted the ReCaptcha free antispam system with a double-word test in 2007, and this test allows the system to formulate a confidence rating for the human by presenting one word the computer does not know with another it does know. People type 200 million captchas globally every day by von Ahn's calculations, while the incredible amount of time people spend playing games drove the CMU professor to initiate a project to tap this pastime to tackle major computational challenges. One game borne from that project, the ESP Game, was designed to enhance Web search using image labeling by asking two randomly paired people on different systems to describe the same image without any communication, and to predict the same word for the image within a time limit. Von Ahn and a group of CMU computer scientists have rolled out four new games to address different challenges in the field of artificial intelligence partly due to the success of the ESP Game.

http://news.cnet.com/8301-1023_3-9989480-93.html

---

© Copyright 2008 Information, Inc. This service may be reproduced for internal distribution.