How to detected misspelled words?

[Iforgotmyoldone]

I'm making a pickit that uses a OCR'er that works 95%+ but it tends to mis letters.

I.e It should be

+8 Intelligence

3% life steal

but it'll have (due to a background image):

+8 Intell gen e

3% li e stea

How would I make a script to detected the misspelled words? What commands should I look up/use.

Exp: I want all items with Intelligence to be picked but if theirs a word/line within 3 letters of it the item it'll be picked up aswell.

[martin]

I'm making a pickit that uses a OCR'er that works 95%+ but it tends to mis letters.

I.e It should be

+8 Intelligence

3% life steal

but it'll have (due to a background image):

+8 Intell gen e

3% li e stea

How would I make a script to detected the misspelled words? What commands should I look up/use.

Exp: I want all items with Intelligence to be picked but if theirs a word/line within 3 letters of it the item it'll be picked up aswell.

That looks difficult!

This is how I would approach it (after not a lot of thought I should warn you)

STEP 1

replace all spaces with '?' and split a line up into an array of words which end with '?'

STEP 2

Loop through the list of word, building up a new word by adding the next word but ignoring words that contain non-alpha characters. If the nex word read starts with '?' then add one letter at a time instead of all because one of the question marks could correctly be a space. At every step see if the word is in the dictionary .

EG

?Intell or Intell

?Intell?gen or Intell?gen

?Intell?gen?e or Intell?gen?e = match found

Obviously there will be some extra logic needed but that's the basic approach I would try. You just need a dictionary or use an on-line spell checking method. Whatever method you use I expect that there will still be errors because false fits will be found.

[rudi]

Hi.

How would I make a script to detected the misspelled words? What commands should I look up/use.

Exp: I want all items with Intelligence to be picked but if theirs a word/line within 3 letters of it the item it'll be picked up aswell.

I think it's not worth to code your own spell checker. Spell checking isn't trivial. So:

95% due to *DROPPED* letters is a really bad result for OCR. Check, if you can tune your OCR with the program's settings (font type, training modes, etc), and if the scanner is giving proper contrast and resolution.

1.) get an OCR (and/or scanner) that work better

2.) after done so, use the *existing* spell checker of your word processing SW (OpenOffice, Word, ...)

Regards, Rudi.