You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@opennlp.apache.org by "Jörn Kottmann (JIRA)" <ji...@apache.org> on 2011/05/24 13:36:47 UTC
[jira] [Closed] (OPENNLP-175) Add support to influence all
available machine learning settings during training of all our components
[ https://issues.apache.org/jira/browse/OPENNLP-175?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Jörn Kottmann closed OPENNLP-175.
---------------------------------
Resolution: Fixed
> Add support to influence all available machine learning settings during training of all our components
> ------------------------------------------------------------------------------------------------------
>
> Key: OPENNLP-175
> URL: https://issues.apache.org/jira/browse/OPENNLP-175
> Project: OpenNLP
> Issue Type: Improvement
> Components: Chunker, Doccat, Maxent, Name Finder, Parser, POS Tagger, Sentence Detector, Tokenizer
> Reporter: Jörn Kottmann
> Assignee: Jörn Kottmann
> Fix For: tools-1.5.2-incubating
>
>
> There are quite some settings which influence the training of a classification model. Currenlty it is difficult to control these settings and many can only be changed through re-compiling. To improve the situation a user should be able to pass an object which contains all the machine learning settings to the various train methods and the command line interface should be extended to accept a properties file which contains all the machine learning settings.
> The settings also contain the training algorithm which enables us to use either maxent or perceptron in all our components.
> The solution which will be implemented was discussed in this thread:
> http://mail-archives.apache.org/mod_mbox/incubator-opennlp-dev/201105.mbox/%3C4DD284DA.1000900@gmail.com%3E
--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira