Hi, Just a note about AD (Portuguese) Chunker Test:
The AD corpus is not separated in train and test parts. So I was splitting it manually until now. I don't like that because I always need to remember the exact size of each part before doing the test. To make things easier now I am using the ChunkerCrossValidator. I did it once with 1.5.1 to get the initial values and replaced the old 1.5.1 values. The test passed. The results are different, but they are due to two changes we did in 1.5.2. I tried reverting the changes and got the same values from 1.5.1, so it is not a regression. Thanks, William On Thu, Oct 27, 2011 at 2:30 AM, James Kosin <[email protected]> wrote: > Jorn, > > The namefinder output hasn't changed in performance from the last > release. Now that I fixed my self inflicted bug on adding tokens for > the parsers... I'm okay. The models are coming out good and are working > correctly. > > One thing... for some reason, when I build, I get a lot of references to > this: > --- > Building jar: C:\Users\James > > Kosin\Documents\NetBeansProjects\thesis\apache\opennlp\opennlp-tools\target\opennlp-tools-1.5.3-incubating-SNAPSHOT-sources.jar > opennlp already added, skipping > opennlp\tools already added, skipping > opennlp\tools\util already added, skipping > META-INF already added, skipping > META-INF\DEPENDENCIES already added, skipping > META-INF\LICENSE already added, skipping > META-INF\NOTICE already added, skipping > > --- > > This happens in several places... and although it really isn't an error > it is annoying. > > James > > On 10/21/2011 9:59 AM, Jörn Kottmann wrote: > > Hi all, > > > > our next release candidate is ready for testing > > > > It can be downloaded from here: > > http://people.apache.org/~joern/releases/opennlp-1.5.2-incubating/rc3/ > > > > To use it in a maven build set the version for opennlp-tools or > > opennlp-uima > > to 1.5.2, and for opennlp-maxent to 3.0.2, and add this URL to your > > settings.xml > > file: > > https://repository.apache.org/content/repositories/orgapacheopennlp-081 > > > > The RC 2 staging repository is dropped now. > > > > The current test plan can be found here: > > https://cwiki.apache.org/OPENNLP/testplan152.html > > > > The release plan can be found here: > > https://cwiki.apache.org/OPENNLP/releaseplanandtasks152.html > > > > Compared to the last RC the following things are fiexed: > > - OPENNLP-327: Doccats bag of word feature generator should not use > > numbers as features > > - OPENNLP-316: Evaluator and CrossValidator programs of the main > > analyzers throw exceptions > > - OPENNLP-317: opennlp.uima.Chunk feature name "type" not allowed > > > > Please pick up items in the test plan and report your results. > > > > Jörn > > > > > > > > > > > >
