Sunday, August 28, 2005

Combatting Spams

In November 1999, there was an online poll conducted to find which is the best graduate school in computer science. But students at Carnegie Mellon wrote a program that voted for around thousand times. The next day, students of MIT also wrote a program that started to vote for them.

How could anyone trust the result of such polls? How would you know if a human has voted or its a program that has voted?

The answer to this is CAPTCHA, an acronym for "Completely Automatic Public Turning Test to tell Computers and Humans Apart". Captcha, coined in 2000, determines whether the user is a human or a program. When signing up for email accounts like Yahoo!, you could have seen a small rectangular graphic containing some distorted words like "none","care","happy" etc. This is known a CAPTCHA process. A human can read these distorted text, but computer programs cannot thus preventing spam and bots from broadcasting unwanted messages to your email.

But still some bots started using Optical character recognition to identify hidden words. The captcha was fist broken by Jitendra Malik. So, to overcome this, people at Palo Alto Research center(PARC) developed something called BaffleText. The process involves printing the image, then scanning it back, thresholding it, adding noise to it and making more worse that even the OCR cannot decipher it.

The main drawback of this technique is that, it is unfriendly to visually impaired people.

No comments: