Incorrect Magazine Article Title
I recently joined Trove for text correcting and all of the other wonderful features of this website and decided to search and correct some Elton John articles. When I did my search one of the results was titled
"ELGON JOHN'S CHRISTMAS CONCERT"
See article: http://trove.nla.gov.au/ndp/del/page...47?zoomLevel=1
(i.e. it should be ELTON). It's understandable how OCR mistook the 'T' for a 'G' given the font Women's Weekly decided to use way back in 1971!
How should one go about alerting to this mistake given I can't change it (or perhaps just don't know how to) because it's the title and not the article text. At the same time does it need changing? Since the article appears in relevant searches because the article contains the correct spelling.
P.S. What an awesome picture of him! And from 1971 - right at the start of his career. I sometimes wish I could travel back to that era - I'm 20 - and in some ways I can with Trove... I love it.
I think you will find the following thread useful in answering your question:
This thread was the result of wondering where the translation for the Title, Sub-title, and Preview of the first four lines come from.
The translated Preview of the first four lines is only supplied for 'articles' and not other sections such as advertising, family notices, lists, etc. However I have just noticed that it seems for articles in the Women's Weekly, which are largely illustrated articles, that they do not supply a translated preview of the first four lines - only a title and sub-title. So the "ELGON JOHN'S Christmas Concert" has been given as the title, and "TV" as the sub-title (i.e the name of the regular segment in this case).
As you will see from the linked thread, the translated Title, Sub-title, and Preview are done manually by the 'vendors' carrying out the computer OCR translation of the whole articles. So therefore the mistake you refer to, "ELGON" instead of "ELTON", is a misreading my a person and not a computer.
Normally the computer OCR translation also includes the title and sub-title (often less accurately translated than the manually supplied version) along with the remainder of the article, so you can correct title there. But what has happened in the case of this article is that the 'fancy' font used for "ELTON JOHN'S" in the original has not been recognised by the computer as text, and so has not been included in the translation. This problem is also very common in advertisements in newspapers, which often use all sorts of fancy fonts, widely spaced letters, etc.
So in this case you have a missing line in the translation. The usual advice here is to include the missing line at the end of the preceding line in the translation box, however in this case because the missing line is the first line, then you will have to include it at the start of the following line. So the first line of correction will become "ELTON JOHN'S Christmas".
This will also be addressing another problem in translation that you will probably find quite commonly in a magazine, as opposed to a newspaper. That is that the title and/or sub-title will often be spread across two pages and there is no way the computer OCR translation is going to pick that up. So the sensible thing to do would be to put the word(s) lost onto the other page back into the line where they would make sense in the text translation. In this case if you were to add "ELTON" and "JOHN'S" as translations on separate pages then anyone searching for "Elton John's christmas" would never find this article unless that exact phrase was used elsewhere in the article. I presume for the majority of articles this will not be a problem because the title supplied by the 'vendor' will be accurate.
The problem of words being separated sometimes occurs in translation of advertisements as well. Particularly where 'drop-cap' letters (i.e. large, over-sized first letter of a word) are used at the start an advertisement - this often results in the first letter being on a separate line to the remainder of the line, and sometimes they end up on lines many lines apart, in a computer OCR translation. I sought advise on what to do about that as well, see the following link:
Another approach to address the title being split across two pages (and not properly translated by the 'vendor') would be to enter the full correct title into a comment box. When anyone conducts a search for a word or phrase it not only searches across the OCR translated text, but also any tags or comments, and the (vendor supplied) titles, sub-titles and preview lines.
If you haven't read them already there is general guidance provided in the Text Correction Guidelines:
There are also some further thoughts on the Guidelines and correction methodology in this thread:
The important thing to remember is that you should correct to match the original, but need to bear in mind that the highest priority should be given to making sure the text is 'findable'.
As I have stated in the link above, the main aims of text correction (in order of importance) should be:
1 - To increase the effectiveness of searching.
2 - To maintain 'textual accuracy' (except where it may cause problems with searching).
3 - To produce a text translation that is suitable for export and use in other documents with a minimum of formatting and 're-correcting' (using various software).
I hope that clears it all up for you.
Last edited by Spearth; 14-08-2011 at 09:05 PM.