[ https://issues.apache.org/jira/browse/OPENNLP-986?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15865791#comment-15865791 ]
ASF GitHub Bot commented on OPENNLP-986: ---------------------------------------- GitHub user tteofili opened a pull request: https://github.com/apache/opennlp/pull/124 OPENNLP-986 - Stupid Backoff as default LM discounting You can merge this pull request into a Git repository by running: $ git pull https://github.com/tteofili/opennlp OPENNLP-986 Alternatively you can review and apply these changes as the patch at: https://github.com/apache/opennlp/pull/124.patch To close this pull request, make a commit to your master/trunk branch with (at least) the following in the commit message: This closes #124 ---- commit 4e566ff62c2f7106bcbc0f9f4f765ceb6040ca3b Author: Tommaso Teofili <teof...@adobe.com> Date: 2017-02-14T13:49:09Z OPENNLP-986 - Stupid Backoff as default LM discounting ---- > Use stupid backoff by default in NGramLanguageModel > --------------------------------------------------- > > Key: OPENNLP-986 > URL: https://issues.apache.org/jira/browse/OPENNLP-986 > Project: OpenNLP > Issue Type: Improvement > Reporter: Tommaso Teofili > Assignee: Tommaso Teofili > Fix For: 1.8.0 > > > {{NGramLanguageModel}} is already using [Stupid > Backoff|http://www.aclweb.org/anthology/D07-1090.pdf] discounting when it > contains more than 1M ngrams. > However since the not very good performance of Laplace smoothing for smaller > models, it'd be better to simply use Stupid Backoff in all cases. -- This message was sent by Atlassian JIRA (v6.3.15#6346)