Yes, the issue is here: https://issues.apache.org/jira/browse/OPENNLP-869
I think we can fix that for the upcoming 1.6.1 release. Jörn On Thu, Oct 20, 2016 at 1:57 AM, Finan, Sean < sean.fi...@childrens.harvard.edu> wrote: > Sure - can you do that? > > Thanks, > Sean > > -----Original Message----- > From: Joern Kottmann [mailto:kottm...@gmail.com] > Sent: Wednesday, October 19, 2016 7:52 PM > To: dev@ctakes.apache.org > Subject: Re: cTAKES Couldn't find parse for: > > This can easily be fixed in the OpenNLP Parser. Should we open an issue > for it? > > Jörn > > On Wed, Aug 24, 2016 at 7:44 PM, Miller, Timothy < > timothy.mil...@childrens.harvard.edu> wrote: > > > IIRC, the "couldn't find parse" message is internal to opennlp so we > > weren't able to turn it off, but then we create a simple flat analysis > > using POS tags for sentences with parse failures. You may need to > > verify this but that is to the best of my recollection. > > Tim > > > > ________________________________________ > > From: Tomasz Oliwa <ol...@uchicago.edu> > > Sent: Wednesday, August 17, 2016 3:01 PM > > To: dev@ctakes.apache.org > > Subject: cTAKES Couldn't find parse for: > > > > Hi, > > > > I am sometimes seeing the "Couldn't find parse for" message in the > > terminal from cTAKES, I assume it is from the MaxentParserWrapper. > > > > However, in the CVD I can still see the TreebankNode entries with a > > tree and nodes with a grammatical nodeType in these cases. How is this > possible? > > What actually does happen when "Couldn't find parse for" occurs? What > > entries in the CAS do not get populated? > > > > You can try it with the following sentence in the CVD that triggers a > > "Couldn't find parse for": > > > > ". . . . . . . . .x . . .Claim . ,. . ;, ;, ; ,.;, . . .Claim;,.;, > > Claim;, .; ,.; ,.; . ., Claim., ,.,;" > > > > Regards, > > Tomasz > > >
