[jira] [Commented] (OPENNLP-641) Add sequence classification support

["Joern Kottmann (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/OPENNLP-641?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13920970#comment-13920970 ] 

Joern Kottmann commented on OPENNLP-641:
----------------------------------------

The beam size parameters can now be specified during training time. We still need to see how we maintain backward compatibility with users, who set it manually before.

I suggest that if the user trained a MaxentModel, then the passed in beam size parameter will be used, if the model, doesn't contain a beam size parameter.

All constructors taking the beam size should be deprecated.

> Add sequence classification support
> -----------------------------------
>
>                 Key: OPENNLP-641
>                 URL: https://issues.apache.org/jira/browse/OPENNLP-641
>             Project: OpenNLP
>          Issue Type: Improvement
>          Components: Machine Learning
>            Reporter: Joern Kottmann
>            Assignee: Joern Kottmann
>             Fix For: 1.6.0
>
>
> The current machine learning integration doesn't support sequence models. That are models which classify an entire sequence at once compared to the classification support we have right now which can only make a single decision at a time.



--
This message was sent by Atlassian JIRA
(v6.2#6252)