Are you digitizing books and don't even know it?

mogui · 05-25-2007, 09:20 PM

Luis von Ahn is conducting a virtual "barn raising" to get free labor from you! Every time you type in one of those CAPTCHAs, the wiggly words used to authenticate you when you sign on to a site, you may be correcting a piece of digitized text. The CAPTCHAs are a type of Turing test. Luis von Ahn won a MacArthur genius award for his idea. Thanks to him, we may have ever more eBooks!

nekokami · 05-26-2007, 10:10 AM

This got a recent BoingBoing mention, too.

NatCh · 05-26-2007, 10:54 AM

It really is quite clever, isn't it? I greatly admire clever.

Bob Russell · 05-26-2007, 11:27 AM

This is really interesting! I wonder if the amount of progress is reasonable. I assume the computer can also tell which bits need human help, so only a small percentage of the text actually uses the CAPTCHA input.

On the other hand, to be able to confirm the turing test results against known interpretation and also get new interpretation from the human, aren't you asking for more input than you would have to otherwise. It sounds like it is a regular turing test PLUS additional work to digitize the books. In other words, the trick seems to be that you have a captive audience to get more effort, not that you use existing effort.

So maybe they should be called "GOTCHAs"?!

nekokami · 05-28-2007, 12:30 AM

Yes, it uses two captchas (or gotchas

), one standard and one for OCR content, and if you get the known one right, it uses the OCR one.

05-25-2007, 09:20 PM	#1
mogui eNigma Posts: 503 Karma: 1335 Join Date: Dec 2006 Location: The Philippines Device: HTC G1 Android FBReader	Are you digitizing books and don't even know it? Luis von Ahn is conducting a virtual "barn raising" to get free labor from you! Every time you type in one of those CAPTCHAs, the wiggly words used to authenticate you when you sign on to a site, you may be correcting a piece of digitized text. The CAPTCHAs are a type of Turing test. Luis von Ahn won a MacArthur genius award for his idea. Thanks to him, we may have ever more eBooks! Last edited by mogui; 05-25-2007 at 09:26 PM.

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Two free books (kobo) from Francesca Lia Block [BOOKS DELETED BY AUTHOR]	koland	Deals and Resources (No Self-Promotion or Affiliate Links)	11	04-08-2010 06:03 AM
Digitizing a book best practices	Linus	Workshop	1	07-13-2009 01:00 PM
Digitizing back covers.	JasonB	News	17	02-15-2009 11:26 AM
Desperately Seeking Software for Digitizing Books.	harryE123	Reading and Management	8	12-17-2008 08:33 PM
Digitizing physical ebooks : hardware recommandation ?	guylhem	Workshop	6	03-05-2008 08:31 AM

05-26-2007, 10:10 AM	#2
nekokami fruminous edugeek Posts: 6,745 Karma: 551260 Join Date: Oct 2006 Location: Northeast US Device: iPad, eBw 1150	This got a recent BoingBoing mention, too.

05-26-2007, 10:54 AM	#3
NatCh Gizmologist Posts: 11,615 Karma: 929550 Join Date: Jan 2006 Location: Republic of Texas Embassy at Jackson, TN Device: Pocketbook Touch HD3	It really is quite clever, isn't it? I greatly admire clever.

05-26-2007, 11:27 AM	#4
Bob Russell Recovering Gadget Addict Posts: 5,381 Karma: 676161 Join Date: May 2004 Location: Pittsburgh, PA Device: iPad	This is really interesting! I wonder if the amount of progress is reasonable. I assume the computer can also tell which bits need human help, so only a small percentage of the text actually uses the CAPTCHA input. On the other hand, to be able to confirm the turing test results against known interpretation and also get new interpretation from the human, aren't you asking for more input than you would have to otherwise. It sounds like it is a regular turing test PLUS additional work to digitize the books. In other words, the trick seems to be that you have a captive audience to get more effort, not that you use existing effort. So maybe they should be called "GOTCHAs"?!

05-28-2007, 12:30 AM	#5
nekokami fruminous edugeek Posts: 6,745 Karma: 551260 Join Date: Oct 2006 Location: Northeast US Device: iPad, eBw 1150	Yes, it uses two captchas (or gotchas ), one standard and one for OCR content, and if you get the known one right, it uses the OCR one.

Advert

Advert