PS: I forwarded Jim's message to one of the Belarusian Wikisourcers
On Tue, Aug 12, 2014 at 11:12 PM, Jim O'Regan joregan@gmail.com wrote:
On 12 August 2014 17:25, Nick White nick.white@durham.ac.uk wrote:
Dear Wikisourcerers,
It's good to hear from you. Wikisource is awesome, as far as I am concerned.
One of the most serious issues was raised by the Belarusian community which
uses 2
different scripts with no commercial OCR support. This means that the volunteers have to type each word manually. We wondered if it would be
possible
to train Tesseract to recognize these old texts using the text that has
been
already typed.
Actually, Tesseract should already have support for Russian and Belarussian "out of the box"; see the 'rus' and 'bel' training data.
'bel' contains Cyrillic; there is also a Latin script ('Łacinka') for Belarusian. (Russian is widely spoken in Belarus, but Russian texts would be added to the Russian Wikisource).
The question I'd have for the Belarusian Wikisourcers is: can they be treated as having an exact mapping? (It doesn't need to be 1:1, I'm aware that, e.g., 'нь' maps to 'ń'). I ask because, as I remember it, there's very little text in Łacinka, and adapting Cyrillic material could be useful.
One thing that wikisource could potentially do for us would be provide loads of proofread, freely reusable "ground truth" data to test Tesseract with. Are there programatic ways of getting at the data, for example downloading all page images and corresponding text that is marked as green, for a specific language / script?
They're all added to a category, so that part should be pretty easy.
-- <Sefam> Are any of the mentors around? <jimregan> yes, they're the ones trolling you
wikisource-l@lists.wikimedia.org