[OPENNLP-1459] Investigate eval-tests failures in OpenNLP 2.1.1 RC1

• Type: Task
• Status: Closed
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 2.1.1
• Fix Version/s: 2.1.1
• Assignee: Jeff Zemerick
• Reporter: Jeff Zemerick
• Created: 30/Jan/23 14:00
• Updated: 31/Jan/23 13:49
• Resolved: 31/Jan/23 13:49

Description

Investigate eval-tests failures in OpenNLP 2.1.1 RC1. Both of these tests pass in OpenNLP 2.1.0.

[ERROR] Failures:
[ERROR]   ArvoresDeitadasEval.evalPortugueseChunkerQn:211->chunkerCrossEval:140 expected: <0.9648211936491359> but was: <0.9651009811896799>
[ERROR]   SourceForgeModelEval.evalChunkerModel:345 expected: <226003515785585284478071030961407561943> but was: <304922886851384639120257052245406261332>

$ mvn -v
Apache Maven 3.8.4 (9b656c72d54e5bacbed989b64718c159fe39b537)
Maven home: /opt/apache-maven
Java version: 11.0.17, vendor: Ubuntu, runtime: /usr/lib/jvm/java-11-openjdk-amd64
Default locale: en_US, platform encoding: UTF-8
OS name: "linux", version: "5.15.0-58-generic", arch: "amd64", family: "unix"

Issue Links

• is caused by: OPENNLP-1332 Default chunker context generator has inconsistent format