Thank you James - great explanation and information. Hi David, There is a new version of the dictionary creator gui distributed with ctakes 4.0. It is 99% the same as the version in sandbox. One important difference is that it now produces a database in hsqldb 2.3.4 - compatible with ctakes 4.0. Another difference (important to you) is that there is an exclusion list in a data file that lists cui / term combinations that can be excluded. By default the medications "toDAY" and "ToMORROW" are in that list. It is obviously done to prevent the frequent false positives that you spoke of in your Amia presentation. For non-vet use these exclusions are pretty valid. Since you are interested in a vet dictionary fi you use the new dictionary creator you can decide whether or not you want them to be included.
Sean -----Original Message----- From: James Masanz [mailto:masanz.ja...@gmail.com] Sent: Saturday, April 22, 2017 11:32 PM To: dev@ctakes.apache.org Subject: Re: cTAKES 4.0rc3 - custom dictionary needs rebuilding? Dave, It's an oversight that we didn't document that. Thanks for pointing that out! I'll update the documentation tomorrow and post again with proper details of the command to use. The short answer is that hsqldb jar includes a GUI. You use the 2.3.4 version of that jar, which is included with the cTAKES 4.0 convenience binary, to open your database. Then you close (shutdown) your DB and it gets converted to 2.3.4 for you. But if the .properties file for your DB indicates it's readonly, you need to have edited the properties file first to remove that. That gives you the idea. I'll write it up properly tomorrow. Thanks for testing, -- James *Sent from my phone.* On Apr 22, 2017 9:39 PM, "David Kincaid" <kincaid.d...@gmail.com> wrote: I finally had some time this weekend to give 4.0.0rc3 a run. The install and configuration instructions worked perfectly on my Linux laptop and I was able to use CVD to run the clinical pipeline successfully against some of the clinical notes I have here. However, I ran into a problem when I tried running my own customized pipeline that includes a custom dictionary that I create (to include the SNOMED veterinary extension). I get an exception while it's loading that the dictionary is not the correct version. So, I assume I just need to recreate that custom database? Or is there a migration utility from the old database version to the new version? I don't remember seeing mention of this, but certainly could have missed it. If it's not in the release notes or install/upgrade instructions it probably should be there. Thanks to everyone for pulling this release together and getting it out the door! - Dave
