From: Kevin Patrick Scannell <scannell@sl...> - 2005-11-28 00:36:36
On 19:34 Sun 27 Nov , Jonathon Blake wrote:
> Does anybody have any pointers for determining whether or not word is
> misspelled, when there are no known e-text dictionaries for the
What language? The best I can offer is to train the
crubadan web crawler on some sample text (if it's not one of the
languages already done here): http://borel.slu.edu/crubadan/stadas.html
Then it can flag "suspicious-looking" words (words containing
improbable three-character sequences as determined by
statistics gathered from the training text). Of course
this doesn't work very well as a reliable spellchecker -
e.g. "tould" looks like perfectly good English from
this perspective, while "syzygy" would surely be flagged
despite being correctly spelled (I think)...
Get latest updates about Open Source Projects, Conferences and News.