[
https://issues.apache.org/jira/browse/OPENNLP-1589?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17864262#comment-17864262
]
ASF GitHub Bot commented on OPENNLP-1589:
-----------------------------------------
rzo1 commented on PR #635:
URL: https://github.com/apache/opennlp/pull/635#issuecomment-2218092187
We found, that the == to equals results in different f-measures.
Guess, we would need to compute baseline f-measure for each test without a
cache and compare those results against the f-measure computed with a cache in
place.
Don't know how the original values we're computed.
> Fix incorrect array check in CachedFeatureGenerator
> ---------------------------------------------------
>
> Key: OPENNLP-1589
> URL: https://issues.apache.org/jira/browse/OPENNLP-1589
> Project: OpenNLP
> Issue Type: Bug
> Components: Name Finder
> Affects Versions: 2.3.3
> Reporter: Martin Wiesner
> Assignee: Richard Zowalla
> Priority: Major
> Fix For: 2.4.0
>
>
> There is an invalid comparison for equality of two arrays in
> CachedFeatureGenerator#createFeatures(..) in line 58.
> Currently, many situations exists in which no improvements by the caching are
> achieved (as intended).
> This should be repaired and a test shall show the cache mechanism is working
> correctly.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
