Huh. Now that's interesting.
May. 26th, 2005 11:29 pm![[personal profile]](https://www.dreamwidth.org/img/silk/identity/user.png)
Aoccdrnig to rsereach at Cmabrigde Uinervtisy, it deosn't mttaer inwaht oredr the ltteers in a wrod are, the olny iprmoatnt tihng is taht the frist and lsat ltteer be in the rghit pclae. The rset can be a taotl mses and you can sitll raed it wouthit a porbelm. Tihs is bcuseae the huamn mnid deos not raed ervey lteter by istlef, but the wrod as a wlohe. Amzanig huh? Yaeh and I awlyas thought
Still, I think it requires context. I mean, given:
iprmoatnt
one doesn't think of important. I know I think of moats. and IP addresses.
But that'd be the first level of recognition... Hmmm. Assigning each word two values. One: length of the word. Two: sum of each character, as a number. Put all the words into a 3D matrix.. rows as lengths of the word, columns as the sum. Then from there, match.
So if we could teach our OCR (and voice recognition systems) grammar, we'd get better results. But that would require the systems to have knowledge of each word. That'd be the second level of recognition.
The third level would be common sense. That's going to be hard, considering how little of it's around. But just so it doesn't come up with "That tree humped." or something equally absurd. Of course, it depends on context... so...
Still, I think it requires context. I mean, given:
iprmoatnt
one doesn't think of important. I know I think of moats. and IP addresses.
But that'd be the first level of recognition... Hmmm. Assigning each word two values. One: length of the word. Two: sum of each character, as a number. Put all the words into a 3D matrix.. rows as lengths of the word, columns as the sum. Then from there, match.
So if we could teach our OCR (and voice recognition systems) grammar, we'd get better results. But that would require the systems to have knowledge of each word. That'd be the second level of recognition.
The third level would be common sense. That's going to be hard, considering how little of it's around. But just so it doesn't come up with "That tree humped." or something equally absurd. Of course, it depends on context... so...
no subject
Date: 2005-05-27 03:42 am (UTC)I think this trick works because the brain takes shortcuts when reading familiar language. Instead of reading letters one by one, it takes it all as a clump and picks out a few 'key' letters to id it. This is enhanced in an environment wrought with spelling errors, like the internte, wher e worsd are frequentlyy juts slightly scrwwed up.
no subject
Date: 2005-05-27 04:50 am (UTC)But... sorchtuts? What do you make of that? And flimaair? Or luggaane?
Mmm, I think I'll keep going. This is fun~ itneasd? echnnaed? enemnorivnt? fletnuqery?
no subject
Date: 2005-05-27 04:55 am (UTC)instead. environment. frequently.
These take longer because the letter associatons the brain expects have been obliterated and patterns the brain is trained to recognize have been created.In the first, "ch", then "lim" and "air", then "lug" and letter repitition.
Each took about 5-10 seconds of staring, but the brain gets it.
no subject
Date: 2005-05-27 05:15 am (UTC)no subject
Date: 2005-05-27 05:16 am (UTC)no subject
Date: 2005-05-27 07:11 pm (UTC)My Brain: three words on first line, so three words in second row. Count: 1st, second-from-last, last.
Yayfor brain hacking.
no subject
Date: 2005-05-27 03:43 am (UTC)no subject
Date: 2005-05-27 04:43 am (UTC)I see withe betewen your liens.
Mm, an urban myth... Meh, nah, it's not important enough to merit researching.
no subject
Date: 2005-05-27 01:12 pm (UTC)no subject
Date: 2005-05-27 03:47 am (UTC)no subject
Date: 2005-05-27 04:51 am (UTC)'cause I'm having fun scrambling words... hmitflae?
no subject
Date: 2005-05-27 07:27 pm (UTC)no subject
Date: 2005-05-27 07:20 am (UTC)Of course, it sucks when you try to do this with numbers.
no subject
Date: 2005-05-27 09:54 am (UTC)but one of the ways that i read is by "word shapes"--the atcaul shape of the word. that's why when some people see it written out they can know it's mispelled, but not know how to fix it.
no subject
Date: 2005-05-27 03:49 pm (UTC)