Joern, Regarding the deprecation of GIS, I noticed that OpenNLP had been refactored between 1.5.2 and 1.6.0 to move the constant *MAXENT_VALUE <https://opennlp.apache.org/documentation/1.6.0/apidocs/opennlp-tools/opennlp/tools/ml/maxent/GIS.html#MAXENT_VALUE>* (that is used to select a MaxEnt classifier) to the GIS class from TrainUtil ( *opennlp.model.TrainUtil*).
Since MAXENT_VALUE had been moved to GIS, I put the corresponding constant for the Naive Bayes classifier (*NAIVE_BAYES_VALUE*) into the NaiveBayesTrainer class. So, if you're moving constants back to something like the TrainUtil, you may want to move the NAIVE_BAYES_VALUE over too. Cohan Sujay Carlos On Mon, Nov 7, 2016 at 2:44 PM, Joern Kottmann <kottm...@gmail.com> wrote: > I also opened a Jira now for the refactoring effort: > https://issues.apache.org/jira/browse/OPENNLP-880 > > Jörn > > On Sun, Nov 6, 2016 at 4:41 PM, Joern Kottmann <kottm...@gmail.com> wrote: > > > On Fri, 2016-11-04 at 12:58 +0000, Russ, Daniel (NIH/CIT) [E] wrote: > > > Hi Jörn, > > > Currently, GIS is a wrapper for the GISTrainer, which ironically > > > is NOT an EventTrainer. There are several ways to handle this. > > > > > > First, don’t touch GIS. This is the class that people know best. It > > > is easy to use if you use the parameter hard coded in the class. We > > > can make GISTrainer an Event trainer and refactor it appropriately to > > > handle parameters in the map. > > > > > > Second, refactor GIS, but keep the static methods in the code. This > > > allows user to keep their code working. > > > > > > Third, refactor GIS and deprecate the static methods. > > > > > > I don’t see any value to the GIS class other than an Adapter. My > > > vote would be the fix the original problem of GISTrainer NOT being an > > > event trainer. > > > > > > I think we are lucky that GISTrainer is not a public class, so that > > allows us to modify it as we need. I agree with you, we should make it > > an EventTrainer. > > > > The GIS class should probably just be deprecated and then removed > > later. > > > > Regards, > > Jörn > > >
