Someone can correct me but as far as I know there is no fuzzy match in the sense I think you mean - a misspelling or bad OCR result (for example 1 replacement character or 1 deleted character).
The closest thing like that that I can think of is the way the LVG component allows cTAKES to consider lexical variants (run vs runs vs ran), but that does not include misspellings, that is for valid variations of the word. -- James On Thu, Jan 25, 2018 at 3:22 PM, Schenk, Gundolf <[email protected]> wrote: > Hi, > > > > Me too would like to have a fuzzy matching in cTAKES. Do we have anything > like it? Or is the typical approach to use a preprocessor to correct for > misspellings, clean up/standardize interpunction and such? > > Would be great to read any thoughts on this. > > > > Cheers, > > Gundolf. > > > > > > *From: *Desteny Child <[email protected]> > *Reply-To: *"[email protected]" <[email protected]> > *Date: *Wednesday, January 24, 2018 at 8:10 AM > *To: *"[email protected]" <[email protected]> > *Subject: *cTAKES fuzzy string matching in NER tasks > > > > Hello, > > I'd like to use cTAKES 4 for Named-Entity Recognition (NER) tasks. Right > now I'm just wondering is it possible to configure cTAKES to use fuzzy > string matching for NER tasks. I need it because very often my documents > are not of the very well quality and the words inside can be a little bit > corrupted (for example after the OCR process). > > If so, could you please point me to the appropriate cTAKES documentation > and samples in order to do it? > > Thanks in advance, > > Mike >
