|
|
We love Captchas!
|
|
|
|
Mac Elite
Join Date: Jan 2004
Status:
Offline
|
|
http://news.bbc.co.uk/1/hi/technology/7567692.stm
Converting scanned text from old books into digital format by handing out the words for use in captchas. Sort of like a human OCR but in reverse. With over 100 million captchas used every day, that's a lot of old text digitised for free.
Inspired.
|
|
|
|
|
|
|
|
|
Addicted to MacNN
Join Date: Nov 2007
Location: In the hearts and minds of MacNNers
Status:
Offline
|
|
|
|
|
|
|
|
|
|
|
Clinically Insane
Join Date: Jul 2005
Location: Vacation.
Status:
Offline
|
|
Maybe I'm being ultra-dumb here, but if the system running the captcha doesn't know what text is being displayed in the graphic then how exactly does it function as a captcha? And if it does know what text the graphic is displaying, then what's the point in the conversion at user-level?
|
Been inclined to wander... off the beaten track.
That's where there's thunder... and the wind shouts back.
|
|
|
|
|
|
|
|
Senior User
Join Date: Nov 2003
Status:
Offline
|
|
I was thinking the same thing, but I guess it runs the same captcha past multiple users, compares all entered words and then picks the one with the most matches.
(does that make any sense?)
(
Last edited by moep; Aug 20, 2008 at 02:25 AM.
)
|
"The road to success is dotted with the most tempting parking spaces."
|
|
|
|
|
|
|
|
Mac Elite
Join Date: Jan 2004
Status:
Offline
|
|
Originally Posted by moep
I was thinking the same thing, but I guess it runs the same captcha past multiple users, compares at all entered words and then picks the one with the most matches.
(does that make any sense?)
Yes that's exactly what they do. With 99.1% accuracy apparently, which is he same a a proffessional human transcriber.
|
|
|
|
|
|
|
|
|
Clinically Insane
Join Date: Jul 2005
Location: Vacation.
Status:
Offline
|
|
Originally Posted by Andrew Stephens
Yes that's exactly what they do. With 99.1% accuracy apparently, which is he same a a proffessional human transcriber.
But how does that captcha the first user to see a particular image?
|
Been inclined to wander... off the beaten track.
That's where there's thunder... and the wind shouts back.
|
|
|
|
|
|
|
|
Clinically Insane
Join Date: Apr 2007
Location: Iowa, how long can this be? Does it really ruin the left column spacing?
Status:
Offline
|
|
That reminds me of another captcha use. A program will take a captcha that blocks automated email signups, display it to a human with the promise of porn if entered correctly, then uses it to automatically sign up for an email account.
|
|
|
|
|
|
|
|
|
Professional Poster
Join Date: Mar 2002
Location: adequate, thanks.
Status:
Offline
|
|
Now that is one clever use of technology.
|
|
|
|
|
|
|
|
|
Mac Enthusiast
Join Date: Nov 2002
Location: More Cowbell...
Status:
Offline
|
|
Originally Posted by Doofy
But how does that captcha the first user to see a particular image?
The captcha system requires the user to decipher two words- one of the two words is known (and is what is used by the client system to prove you are human) the other is an unknown from a scanned text, and your response is added to a database. You dont know which is known and which is unknown, so you answer both.
|
|
|
|
|
|
|
|
|
Moderator Emeritus
Join Date: Apr 2001
Location: Up In The Air
Status:
Offline
|
|
Captchas are usually a bad idea for security- the email and blogging systems that use captchas don't circumvent spamming, and do inhibit the sight-disabled.
At least using them for OCR serves a decent purpose. Using them for security is a waste of time.
|
|
|
|
|
|
|
|
|
Clinically Insane
Join Date: Jul 2005
Location: Vacation.
Status:
Offline
|
|
Originally Posted by MarkLT1
The captcha system requires the user to decipher two words- one of the two words is known (and is what is used by the client system to prove you are human) the other is an unknown from a scanned text, and your response is added to a database. You dont know which is known and which is unknown, so you answer both.
Ahhhh. That makes sense now. Thanks.
|
Been inclined to wander... off the beaten track.
That's where there's thunder... and the wind shouts back.
|
|
|
|
|
|
|
|
Banned
Join Date: Jun 2005
Location: Indy.
Status:
Offline
|
|
Interesting.
I got an eye raising captcha when signing up for coupons for digital converter boxes this morning:
|
|
|
|
|
|
|
|
|
Clinically Insane
Join Date: Jun 2001
Location: planning a comeback !
Status:
Offline
|
|
Why the fark do they not use OCR ?
This is the absolute mostest stupidest idea I have heard today.
(Yes, only today. Thanks to my employer, I run across a lot of stupid sh!t).
-t
|
|
|
|
|
|
|
|
|
hayesk
|
|
Originally Posted by turtle777
Why the fark do they not use OCR ?
Because it's not good enough on old text. These aren't crisp and clear laser prints they're dealing with.
This is the absolute mostest stupidest idea I have heard today.
(Yes, only today. Thanks to my employer, I run across a lot of stupid sh!t).
You only think it's stupid because you thought OCR was a good solution.
|
|
|
|
|
|
|
|
|
Clinically Insane
Join Date: Jun 2001
Location: planning a comeback !
Status:
Offline
|
|
Originally Posted by hayesk
Because it's not good enough on old text. These aren't crisp and clear laser prints they're dealing with.
You only think it's stupid because you thought OCR was a good solution.
I don't buy this. OCR these days is highly sophisticated. I doesn't even have problems recognizing handwriting w/o training. Example: Evernote: even your hand-scribbled notes will be converted using OCR, no training needed.
Even if a text is barely readable, a computer with OCR can do a much better job trying to understand what the different characters are by cross-comparing to other sections in that book.
-t
|
|
|
|
|
|
|
|
|
Mac Enthusiast
Join Date: Nov 2002
Location: More Cowbell...
Status:
Offline
|
|
Originally Posted by turtle777
I don't buy this. OCR these days is highly sophisticated. I doesn't even have problems recognizing handwriting w/o training. Example: Evernote: even your hand-scribbled notes will be converted using OCR, no training needed.
Even if a text is barely readable, a computer with OCR can do a much better job trying to understand what the different characters are by cross-comparing to other sections in that book.
-t
IIRC, they start with OCR, and it properly encodes most of the text. The OCR software flags words that it can not recognize, and sends it off to be interpreted by the peoples.
|
|
|
|
|
|
|
|
|
Clinically Insane
Join Date: Jun 2001
Location: planning a comeback !
Status:
Offline
|
|
Originally Posted by MarkLT1
IIRC, they start with OCR, and it properly encodes most of the text. The OCR software flags words that it can not recognize, and sends it off to be interpreted by the peoples.
Now THAT would make sense.
-t
|
|
|
|
|
|
|
|
|
Posting Junkie
Join Date: May 2001
Location: Brisbane, Australia
Status:
Offline
|
|
I was wondering why I got a weird captcha containing an indecipherable name consisting of punctuated initials the other day. Something like D.C.tol. I thought I'd never get it right, but it got through. Must be this thing then
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
Forum Rules
|
|
|
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts
|
HTML code is Off
|
|
|
|
|
|
|
|
|
|
|
|