|
In Reply to: Why not start with "The Girl Who Wouldn't" posted by Alan on September 05, 2003 at 14:03:44:
"I've been trying to encourage exfamily.org to post whatever Mo letters they have, but for some reason, they are procrastinating."
I'm sure the administrators of this site have good reasons for not doing what seems to be an easy thing. They've looked into all kinds of angles I'm sure.
"If any of you guys have copies of interesting Mo letters, you could use a scanner and OCR software to convert them into text files, which could then be posted on the net. No need to spend time retyping into a word processor, just use OCR technology to do it for you. Hey, if a blind guy like me can do it, anybody can!"
Well, Alan, you might have access to very good hardware, software and had very optimal conditions for your scanning efforts which made it so easy for you.
Scanning to OCR (Optical Character Recognition) technology is far from perfect.
The problem is that it's not easy to scan multiple columns on a page, as most OCR spftware will read text jumping from one column to the next reading things horizontally in straight lines. You have to use carbon paper or something to block off columns and do them one at a time.
There is also the problem of thin paper being slightly see-through, what is on the other side of it can bleed through and confuse the OCR software. Most Mo letters are printed on very thin paper.
On top of that if the font is not recognizable, or even if the printing isn't picture perfect and deviates in too much, words or characters get messed up or mixed up.
Most of the messes can be fixed with a spell check, but as we know, even spell checks can miss things like context - it happens frequently that the spelling is perfect, but something makes no sense or the document states something completely differrent than the original.
So, even after successful scanning, it still requires detailed A B comparison proofing. All in all, it's still a lot less work than typing, but it ain't a walk in the park.
