Allow users to correct OCR text

[Alex Osborne]

It'd be nice to have an interface that allowed users to correct the OCRed text in a wiki-like manner. The Australian newspaper delivery system beta is trying this out and there are some amazing users who correct tens of thousands of lines of text:

http://ndpbeta.nla.gov.au/ndp/del/hallOfFame

Something to keep in mind is making sure the OCR coordinates still match up with the edited text so we don't lose the ability to do highlighting. The newspapers system handles this by having the edit interface work on a per-line basis (so the line breaks are kept intact) and then matching up edited words with the originals.

[mangtronix (mang)]

FYI the reCAPTCHA project is collecting OCR correction data for some Archive books. We aren't yet incorporating that data back into our books. http://recaptcha.net/