How to use with OCR-ed text? #127
Description
Hello.
I need to process result of OCR. There are 2 problems - addition of dot/comma/dash punctuation (due to noise in
input images) and similarity of some letters in latin and cyrillic alphabets (the text I recognize has a mix of scripts).
How can I modify/prepare input for rapidfuzz for this particular case? I mean, can I specify a sort of weight for character comparision (for example, to specify that 'error'/distance between cyrillic letter O and latin letter O and digit 0 are minimal, and insertion of . or , is counted less then insertion of letters)?
Also, in some cases, my vocabulary is small (a few hundreds words), can this be used somehow to improve score calculation? The speed is not a problem (compared to the speed of OCR). Thanks in advance.