| commit | 66375000f75911b65a5d345fefb0e355a978472d | [log] [tgz] |
|---|---|---|
| author | Jörn Kottmann <joern@apache.org> | Mon Aug 24 21:28:41 2015 +0000 |
| committer | Jörn Kottmann <joern@apache.org> | Mon Aug 24 21:28:41 2015 +0000 |
| tree | 1d180e60a726f6053d72f6c4fc52be53b58e395d | |
| parent | 729117f97c48cdbc700aae7ea81a862814c7456c [diff] |
OPENNLP-807 We have worked on the integration of the existing approaches.
MFS and IMS now work independently, (will make unit tests).
Mostly, we have formatted the IMS approach to be similar to other tools.
IMS now also saves and loads a model file per word for its training data instead of 2 separate files (made as artifacts).
Thanks to Anthony Beylerian for providing a patch.