[
https://issues.apache.org/jira/browse/OPENNLP-1964?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Richard Zowalla updated OPENNLP-1964:
-------------------------------------
Fix Version/s: 3.0.0
> Remove the Japanese add-on
> --------------------------
>
> Key: OPENNLP-1964
> URL: https://issues.apache.org/jira/browse/OPENNLP-1964
> Project: OpenNLP
> Issue Type: Task
> Components: Name Finder
> Reporter: Richard Zowalla
> Assignee: Richard Zowalla
> Priority: Minor
> Fix For: 3.0.0
>
> Time Spent: 20m
> Remaining Estimate: 0h
>
> Remove the {{japanese}} module and its leftover Ant {{build.xml}} from
> opennlp-addons. Most of its feature generators are copies of the ones in
> {{opennlp.tools.util.featuregen}}.
> *Replacement:* core now handles Japanese itself:
> * {{StringPattern}} detects Han, Hiragana and Katakana, and
> {{FeatureGeneratorUtil}} emits the {{jah}}/{{jak}} token classes.
> * {{LatticeTokenizer}} with a MeCab dictionary (OPENNLP-1894) segments the
> text and returns each morpheme with its features, which covers the
> {{word/POS}} tokens {{AuxiliaryInfo*}} expected.
> The module also put classes in {{opennlp.tools.namefind}} and
> {{opennlp.tools.util.featuregen}}, packages that {{opennlp-runtime}} already
> owns.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
