Release v1.0.2 - Bugfix/Maintenance and Scoring Behaviour Change

Supports Elasticsearch v7.3.0

• Various Bugfixes Done in Internal Version that Were not Copied Over Properly are Now Copied Over
• Scoring Behavior Change: Documents with All-Positive Mentions when Search Query Consists of ONLY Negated Terms will now no longer default to a scoring weight of 1.

Elasticsearch 7.3.0, Configuration, and ConText Updates

• Elasticsearch 7.3.0 update
• Updated ConText rules
• Simplified configuration to a yml file
• Removal/Revert of lift-analysis code changes to bring codebase back to full-featureset as originally described

v1.0.1prerelease

Note: this version is untested and may not function correctly. If referencing this repository from the paper submission, please use the initial v1.0.0 release version and make appropriate changes

Untested: user config instead of hardcoded. Revert lift analysis changes

Initial Public Release Version

Initial Release of the Elasticsearch NLP Plugin. Note that this public version differs slightly from the system described in the paper

• Weights are different and hardcoded and certain items are commented out due to an ongoing lift analysis against a differing corpus. To fix this, go to org/ohnlp/elasticsearchnlp/scoring/NLPPayloadScoringWeightFunction.java and make appropriate changes.
• Clinical Stopwords Filtering is enabled by default. To test the system in isolation, as was done in the paper, go to src/main/java/org/ohnlp/elasticsearchnlp/analyzers/NLPAnalyzer.java and comment out line 36. If this is not for testing the system's performance, but rather for use, leave this enabled.