You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@opennlp.apache.org by "Pablo Duboue (KeaText) (JIRA)" <ji...@apache.org> on 2015/04/02 19:27:53 UTC
[jira] [Closed] (OPENNLP-764) UIMA NameFinder Annotator should call
clearAdaptiveData
[ https://issues.apache.org/jira/browse/OPENNLP-764?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Pablo Duboue (KeaText) closed OPENNLP-764.
------------------------------------------
Resolution: Invalid
Further review of the code pointed from Tommaso shows the bug is invalid. Moreover my "fix" resets the adaptive behavior in *each* sentence! Do not use it.
Thanks Tommaso!
> UIMA NameFinder Annotator should call clearAdaptiveData
> -------------------------------------------------------
>
> Key: OPENNLP-764
> URL: https://issues.apache.org/jira/browse/OPENNLP-764
> Project: OpenNLP
> Issue Type: Bug
> Components: UIMA Integration
> Reporter: Pablo Duboue (KeaText)
> Assignee: Tommaso Teofili
> Fix For: 1.6.1
>
>
> Following the documentation (http://opennlp.apache.org/documentation/1.5.3/manual/opennlp.html#tools.namefind.recognition.api) and the discussion in OPENNLP-627, the adaptive data has to be called after each document has been processed.
> This is missing in the UIMA integration, causing severe degradation when processing more than a few thousand documents.
> The following patch fixes it:
> Index: opennlp-uima/src/main/java/opennlp/uima/namefind/NameFinder.java
> ===================================================================
> --- opennlp-uima/src/main/java/opennlp/uima/namefind/NameFinder.java (revision 1670422)
> +++ opennlp-uima/src/main/java/opennlp/uima/namefind/NameFinder.java (working copy)
> @@ -169,6 +169,8 @@
> documentConfidence.add(prob);
> }
>
> + mNameFinder.clearAdaptiveData();
> +
> return names;
> }
>
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)