Badly proofed e-books - Page 5

Sparrow · 08-12-2009, 03:57 AM

Quote:

Originally Posted by Ea

Sometimes the OCR software mangles things so much it's uncomprehensible, or you just can't guess the right word.

If it's a public domain book, you could do a search for the text you have in Google Books - the search hits will return text snippets from their books and hopefully they'll contain the missing word.
It'll return text from Snippet view and No Preview books - which you can't ordinarily access (afaik).
I'm having to do this quite a lot for the book I'm proofing at the moment.

E.g.
The PDF I'm using has "and the dresses of the ladies, ....ped about the piano"

Searching Google Books for the text and book name:
"and the dresses of the ladies" Diana Trelawny
I can see the missing text is "as they stood grouped".

Ea · 08-12-2009, 05:28 AM

Quote:

Originally Posted by Sparrow

If it's a public domain book, you could do a search for the text you have in Google Books - the search hits will return text snippets from their books and hopefully they'll contain the missing word.
It'll return text from Snippet view and No Preview books - which you can't ordinarily access (afaik).

No, it's stuff that I own - but then I have the scan and can display it beside the text I'm working on.

AlexBell · 08-12-2009, 05:40 AM

Quote:

Originally Posted by Ea

On Mac I can do a search and replace of line breaks, so I can remove them - but I haven't been able to find a metod that works in Windows. I remember an article on Lifehacker I read years ago, about reformatting Project Gutenberg books, but I couldn't get any of the suggested methods to work.

I use the inexpensive Atlantis word processor. All one needs to do it select the text from which one wants to remove line breaks, press Ctrl-Shift-U, and the line breaks are gone in seconds. Highly recommended.

Regards, Alex

HarryT · 08-12-2009, 05:49 AM

Quote:

Originally Posted by Ea

Sometimes the OCR software mangles things so much it's uncomprehensible, or you just can't guess the right word.

What's even more insidious is when OCR is combined with a spell checker, and a "mangled" word has been replaced by another word, which "fits in" to the sentence, and yet is totally wrong.

A good example is the Erksine Childers classic spy story "The Riddle of the Sands". Pretty much all the "free" versions of it, PG included, in a paragraph describing the appearance of the cabin of a boat, include a mysterious reference to "banks of yam". I defy anyone to guess what that really should be, without recourse to an original page scan or a printed copy of the book.

The correct text, in case you're wondering, is "hanks of yarn"

.

sony_fox · 08-12-2009, 06:30 AM

Couple of points:

Authors' proofing. I follow the blogs of a big name, make neough money to live off it professional genre fiction authors. They spend days manually proofing the 'correct' galleys returned to them by the publishers for final checks. It is time they budget in the writing of any book, so many days for the writing, plus so many extra for the proofing afterwards. What's worse is that by the time they get the galleys they are already mid-flow in a different story, and have to break mindsets to focus on the 'old' story properly. Hard work indeed but part of the job. One is now re-editing corrupted files of backcatalog to sell from their own webportal as ebooks.

What should I do when I find an error. Once recent purchased and DRMd ebook, was mostly ok. Until about page 500 where for the next 50 pages wordswerefrequently runaltogether it was really quite extremely annonying. SHould I contact the author, the seller or the publisher? or all of them?

HarryT · 08-12-2009, 06:32 AM

Quote:

Originally Posted by sony_fox

What should I do when I find an error. Once recent purchased and DRMd ebook, was mostly ok. Until about page 500 where for the next 50 pages wordswerefrequently runaltogether it was really quite extremely annonying. SHould I contact the author, the seller or the publisher? or all of them?

Contact the seller initially. Reputable sellers will pass the problem report back to the publisher (who are generally the ones who actually create the eBook).

Sparrow · 08-12-2009, 06:38 AM

Quote:

Originally Posted by HarryT

What's even more insidious is when OCR is combined with a spell checker, and a "mangled" word has been replaced by another word, which "fits in" to the sentence, and yet is totally wrong.

A good example is the Erksine Childers classic spy story "The Riddle of the Sands". Pretty much all the "free" versions of it, PG included, in a paragraph describing the appearance of the cabin of a boat, include a mysterious reference to "banks of yam". I defy anyone to guess what that really should be, without recourse to an original page scan or a printed copy of the book.

The correct text, in case you're wondering, is "hanks of yarn"

.

I reckon that would be quite easy to figure out.

Given the context, 'yam' is obviously 'yarn'.
'banks' is more of a puzzler - but 'b' for 'h' is a very common OCR error, and about the only correction candidate that comes to mind for 'banks'.

Ea · 08-12-2009, 07:17 AM

[QUOTE=HarryT;551376]What's even more insidious is when OCR is combined with a spell checker, and a "mangled" word has been replaced by another word, which "fits in" to the sentence, and yet is totally wrong.
...[QUOTE]
Yikes! That is bad. It's hard to spot words that the spell checker doesn't catch.

Quote:

Originally Posted by AlexBell

I use the inexpensive Atlantis word processor. All one needs to do it select the text from which one wants to remove line breaks, press Ctrl-Shift-U, and the line breaks are gone in seconds. Highly recommended.

Regards, Alex

But if you had a text where you want to remove single line breaks and replace double empty lines with true paragraph breaks - can you do that?
(I know there's some difference between line break and paragraph break, but I haven't looked into it, so I might be missing some information)

Lemurion · 08-12-2009, 07:31 AM

I do a lot of freelance editing, proofing, and light rewriting. Almost everything I've seen people complain about (and I've complained about the same things myself) is stuff that's my job, not the author's.

As has been said before, a lot of this is things the author literally cannot see; their brain fills in what's supposed to be there. So when someone tells the author to correct an ebook they're often asking them to do something that's not their job, that they're uniquely ill-suited for, and that they may not even know about because they often have very little if anything to do with ebook releases.

Yes, many authors will do what they can, but by this point it's out of their hands.

slayda · 08-12-2009, 10:27 AM

[QUOTE=Ea;551462][QUOTE=HarryT;551376]What's even more insidious is when OCR is combined with a spell checker, and a "mangled" word has been replaced by another word, which "fits in" to the sentence, and yet is totally wrong.
...

Quote:

Yikes! That is bad. It's hard to spot words that the spell checker doesn't catch.

But if you had a text where you want to remove single line breaks and replace double empty lines with true paragraph breaks - can you do that?
(I know there's some difference between line break and paragraph break, but I haven't looked into it, so I might be missing some information)

In MS Word they call them "Manual line break" and "Paragraph mark". Look in the "Find and Replace" window (from the "Edit" pulldown menu), click "More", then click "Special" to see what you can "find" and "replace".

There is also a "Show/Hide"button to allow you to see the various formating symbols. You can they see the difference between these two marks.

Maybe this will help or maybe I totally missed the problem.

Ea · 08-12-2009, 10:34 AM

Quote:

Originally Posted by slayda

...

In MS Word they call them "Manual line break" and "Paragraph mark". Look in the "Find and Replace" window (from the "Edit" pulldown menu), click "More", then click "Special" to see what you can "find" and "replace".

There is also a "Show/Hide"button to allow you to see the various formating symbols. You can they see the difference between these two marks.

Maybe this will help or maybe I totally missed the problem.

This is great

It's not a current problem on my Mac, but it's been bothering me how to handle it, and I may well get a Windows machine next time.

corroonb · 08-12-2009, 11:35 AM

A trick I've found with OCR errors is to identify the consistent errors and look for other words that might not be picked up with a spell check. Obviously this only work well if the error occurs all the time as you would expect of an automated process.

For example I had an OCR text that had replaced every cl at the start of a word with d. It was easy to find the words like dothes and doset with a spell checker and do a global replace but I had to search for every word that makes sense with a cl and a d in front of it using a dictionary. And you can't use a global replace with dean/clean or dosed/closed as the context has to be checked.

Apologies if this is obvious.

Ea · 08-12-2009, 11:43 AM

Quote:

Originally Posted by corroonb

A trick I've found with OCR errors is to identify the consistent errors and look for other words that might not be picked up with a spell check. Obviously this only work well if the error occurs all the time as you would expect of an automated process.

For example I had an OCR text that had replaced every cl at the start of a word with d. It was easy to find the words like dothes and doset with a spell checker and do a global replace but I had to search for every word that makes sense with a cl and a d in front of it using a dictionary. And you can't use a global replace with dean/clean or dosed/closed as the context has to be checked.

Apologies if this is obvious.

It's a good idea. I've sort of been doing this already, but it's the same as being completely conscious about it, and I've never thought to use a dictionary to help.

starrigger · 08-12-2009, 11:50 AM

Quote:

Originally Posted by HarryT

Contact the seller initially. Reputable sellers will pass the problem report back to the publisher (who are generally the ones who actually create the eBook).

Actually, it is often the seller, not the publisher, who generates the ebook. Contact the seller. FW, Amazon, whoever.

The error might come from the original file. More likely it came from the conversion.

tomsem · 08-12-2009, 12:10 PM

It seems there should be a way to exploit the fact that ebooks are in electronic form, and that we are all connected by the internet. Every reader is a potential proof-reader.

Windows and OS X have a 'crash reporting' mechanism that allows users who experience crashes to send a report to the software publisher in question. Something analogous could be developed for ebooks and built into the reader software (at least for those devices which support annotation). One might even institute a microcredit scheme so that people who report the errors are rewarded in some tangible way. (hmm, publishers could intentionally introduce errors and give credits to the first 100 readers who find it, to encourage this proofing activity..)

So the idea is that users who encounter an error would invoke their reader's 'report an error' function, which would flag the location and allow the user to type a short note as to the nature of the error, the ebook version, the reader's contact info (if they opt in) etc. These error reports would be collected and forwarded or sent directly to the publisher when the device is 'connected' to the internet or tethered to a host computer. The publisher would then resolve the errors, publish a new edition and make it available for download to anyone who owns that title or is purchasing anew. The reader's librarian software could periodically check for and download updates (Amazon already is set up to update titles automatically - maybe a little too automatically in some cases).

eBook marketplaces that institute such a self-correcting system would become preferred to those that do not.

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
The Adventures of Joe Nobody and the Badly Formatted Epub	mklynds	Sigil	44	01-30-2013 02:43 PM
Classic Bought a Badly Formatted Book From B&N	lionel47	Barnes & Noble NOOK	11	05-22-2010 04:31 PM
Unutterably Silly How To Write Badly Well	Madam Broshkina	Lounge	4	11-04-2009 08:26 AM
battery question (I let it drain really badly)	rheostaticsfan	Bookeen	5	11-01-2008 03:21 PM
Bricked iLiad after badly done reflash ?	Pode	iRex	6	05-19-2008 03:42 PM

08-12-2009, 06:30 AM	#65
sony_fox Zealot Posts: 109 Karma: 84 Join Date: Jun 2009 Location: Manchester Device: Kobo Auroa H2O	Couple of points: Authors' proofing. I follow the blogs of a big name, make neough money to live off it professional genre fiction authors. They spend days manually proofing the 'correct' galleys returned to them by the publishers for final checks. It is time they budget in the writing of any book, so many days for the writing, plus so many extra for the proofing afterwards. What's worse is that by the time they get the galleys they are already mid-flow in a different story, and have to break mindsets to focus on the 'old' story properly. Hard work indeed but part of the job. One is now re-editing corrupted files of backcatalog to sell from their own webportal as ebooks. What should I do when I find an error. Once recent purchased and DRMd ebook, was mostly ok. Until about page 500 where for the next 50 pages wordswerefrequently runaltogether it was really quite extremely annonying. SHould I contact the author, the seller or the publisher? or all of them?

08-12-2009, 07:31 AM	#69
Lemurion eReader Posts: 2,750 Karma: 4968470 Join Date: Aug 2007 Device: Note 5; PW3; Nook HD+; ChuWi Hi12; iPad	I do a lot of freelance editing, proofing, and light rewriting. Almost everything I've seen people complain about (and I've complained about the same things myself) is stuff that's my job, not the author's. As has been said before, a lot of this is things the author literally cannot see; their brain fills in what's supposed to be there. So when someone tells the author to correct an ebook they're often asking them to do something that's not their job, that they're uniquely ill-suited for, and that they may not even know about because they often have very little if anything to do with ebook releases. Yes, many authors will do what they can, but by this point it's out of their hands.

08-12-2009, 11:35 AM	#72
corroonb Addict Posts: 317 Karma: 1232685 Join Date: Nov 2008 Location: Ireland Device: Kindle Voyage, Kobo Aura, Nexus 9	A trick I've found with OCR errors is to identify the consistent errors and look for other words that might not be picked up with a spell check. Obviously this only work well if the error occurs all the time as you would expect of an automated process. For example I had an OCR text that had replaced every cl at the start of a word with d. It was easy to find the words like dothes and doset with a spell checker and do a global replace but I had to search for every word that makes sense with a cl and a d in front of it using a dictionary. And you can't use a global replace with dean/clean or dosed/closed as the context has to be checked. Apologies if this is obvious.

08-12-2009, 12:10 PM	#75
tomsem Grand Sorcerer Posts: 6,950 Karma: 27060153 Join Date: Apr 2009 Location: USA Device: iPhone 15PM, Kindle Scribe, iPad mini 6, PocketBook InkPad Color 3	It seems there should be a way to exploit the fact that ebooks are in electronic form, and that we are all connected by the internet. Every reader is a potential proof-reader. Windows and OS X have a 'crash reporting' mechanism that allows users who experience crashes to send a report to the software publisher in question. Something analogous could be developed for ebooks and built into the reader software (at least for those devices which support annotation). One might even institute a microcredit scheme so that people who report the errors are rewarded in some tangible way. (hmm, publishers could intentionally introduce errors and give credits to the first 100 readers who find it, to encourage this proofing activity..) So the idea is that users who encounter an error would invoke their reader's 'report an error' function, which would flag the location and allow the user to type a short note as to the nature of the error, the ebook version, the reader's contact info (if they opt in) etc. These error reports would be collected and forwarded or sent directly to the publisher when the device is 'connected' to the internet or tethered to a host computer. The publisher would then resolve the errors, publish a new edition and make it available for download to anyone who owns that title or is purchasing anew. The reader's librarian software could periodically check for and download updates (Amazon already is set up to update titles automatically - maybe a little too automatically in some cases). eBook marketplaces that institute such a self-correcting system would become preferred to those that do not.

Advert

Advert