Thursday, July 2, 2009

Great NOVA Video - CAPTCHA and reCAPTCHA

You've probably run into squiggly words like the "terrapin" in the figure. It is a kind of reverse "Turing Test" that allows a computer to distinguish a real human from a computer pretending to be a human.

We are great at reading characters that have been distorted to the point they baffle computers. When you sign up for a new email account they make you interpret those characters to be sure you are not a computer program signing up for multiple accounts to be used for sending spam. The technique is called "CAPTCHA" and was invented by Luis von Ahn, see this great NOVA Video about his ingenious work.

Ahn, then a grad student at Carnegie-Mellon and now a professor there, solved another tough problem using a variation of CAPTCHA. Google and others are scanning old books and digitizing the text. The problem is words in old books with obsolete typefaces may be misaligned or smudged or otherwise distorted to the point they can't be reliably read by computers. That is where what Ahn calls "reCAPTCHA" comes in!

The word "legume" in the above figure is from some old book and can't be reliably read by the computer. So, as part of the CAPTCHA process of signing up for an email account, reCAPTCHA presents two words. One is the distorted CAPTCHA word that the computer knows and the other is a word copied from some old book. If the human can correctly interpret the distorted word, they figure he or she can interpret the word from the old book correctly as well. Since millions of people sign up for various computer accounts every day, Ahn has put us all to work helping computers digitize the text of old books!


Ira Glickstein


PS: Thanks to my son-in-law David for putting my wife Vi and me on to this NOVA video!

PPS: In the video, a talking computer is asked: "What is a perfect date?" The answer is "June 23, 1912". I know why. Do you?

1 comment:

Vi Glickstein said...

An amazing rewrite of history by Ira (my husband)who thinks he is the arbiter of the definition of feminism. I find Sarah Palin to be the sneaky, mean-spirited, uneducated, inarticulate person who has risen to the level of her incompetence(e.g. Peter Principle). I found her resignation statement to be a stream of consciousness with her own version of the facts yet again.

She speaks of the politics of personal destruction as if only she is the innocent recipient of this and had nothing to do with the horrors of the recent campaign. I live in Florida and heard the Palin rallies with the hateful statements made by her and her supporters.

Ira has drawn a very narrow model of feminism to me in private conversations regarding a woman who has been elected governor or US Senator: limited money, no family wealth, got herself elected. Well there are examples of women Governors such as Ann Richards (TX) and Ruth Ann Minner (Delaware) and Margaret Chase Smith. But there are more - see http://en.wikipedia.org/wiki/List_of_female_state_governors_in_the_United_States.

Concerning Letterman's comment about A-Rod and Palin's daughter, Letterman's writers just knew Palin was there with A daughter who they assumed was Bristol. The "joke" was not about the younger daughter who was present - it was about the 18yr old. You are just like the other GOP die-hards who insist on twisting everything to your "advantage". Note that all the comments made about Bristol during the campaign by the late-night hosts were never answered. Palin answers them all now and twists them around because it enhances her 15 minutes of fame.

She is a quitter! She could have run out her term and worked at solving the problems her state is experiencing now that oil prices have collapsed from last summer. She has managed to alienate people who have worked for her and with her. She never worked at being governor - never went to Juneau but stayed in Wasilla and billed the state for per diem for the 3 months the legislature is in session. The reasons she is leaving is that her popularity is sinking like a rock (80+% just 9 months ago and now just 50%). She sees a way of making beaucoup money (her book deal which of course will be written by someone else and via speech fees to the conservative die-hards). The lame-duck analogy is a joke - the only part of it that is true is the "lame" part. Regarding the photoshopped photo of Trig, this was done by someone within her state - see http://www.huffingtonpost.com/2009/06/25/palin-attacks-blogger-for_n_220668.html.

Oh well - I could say so much more but I will leave it for now.

Vi Glickstein