[
https://issues.apache.org/jira/browse/OPENNLP-673?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13966477#comment-13966477
]
William Colen commented on OPENNLP-673:
---------------------------------------
The issue is caused by the feature generator events with missing predicates.
The event should have the format:
<pred>=<instance>
For example:
ng=hello,world
> Doccat NGramFeatureGenerator not working
> ----------------------------------------
>
> Key: OPENNLP-673
> URL: https://issues.apache.org/jira/browse/OPENNLP-673
> Project: OpenNLP
> Issue Type: Bug
> Components: Doccat
> Affects Versions: 1.6.0
> Reporter: William Colen
> Assignee: William Colen
>
> Training a document categorizer module using NGramFeatureGenerator always
> fails. All events are dropped.
> (...)
> Dropped event talk.religion.misc:[From:, (...)
> Dropped event talk.religion.misc:[From:, (...)
> done.
> Sorting and merging events...
> Exception in thread "main" java.lang.NullPointerException
> at opennlp.tools.ml.maxent.GISTrainer.trainModel(GISTrainer.java:264)
> at opennlp.tools.ml.maxent.GIS.trainModel(GIS.java:298)
> at opennlp.tools.ml.maxent.GIS.doTrain(GIS.java:83)
> at opennlp.tools.ml.maxent.GIS.doTrain(GIS.java:1)
> at
> opennlp.tools.ml.AbstractEventTrainer.train(AbstractEventTrainer.java:93)
> at opennlp.tools.ml.model.TrainUtil.train(TrainUtil.java:53)
> at
> opennlp.tools.doccat.DocumentCategorizerME.train(DocumentCategorizerME.java:116)
> at
> opennlp.tools.doccat.DoccatCrossValidator.evaluate(DoccatCrossValidator.java:72)
> at
> opennlp.tools.cmdline.doccat.DoccatCrossValidatorTool.run(DoccatCrossValidatorTool.java:99)
> at opennlp.tools.cmdline.CLI.main(CLI.java:224)
> Done indexing.
> Incorporating indexed data for training...
--
This message was sent by Atlassian JIRA
(v6.2#6252)