[OPENNLP-778] Parser is not configured correctly for 1.5.x models

• Type: Bug
• Status: Closed
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 1.6.0
• Fix Version/s: 1.6.0
• Component/s: Parser
• Assignee: Jörn Kottmann
• Reporter: Jörn Kottmann
• Created: 20/May/15 15:20
• Updated: 20/May/15 15:23
• Resolved: 20/May/15 15:23

Description

The Parser in 1.6.0 sets a couple of parameters during training time which are stored inside the model.

Two of those parameters:

• POS Tagger beam size
• Custom chunker sequence validation

Are not set up for 1.5.x models.

To fix this issue the ParserModel should be modified to detect when a 1.5.x model is loaded and then inject those settings into the model.