constit parser desc
The model in parsermodel is trained on Mipacq data (clinical notes, clinical questions, medpedia, etc.) as well as the Penn Treebank (Wall Street Journal). This model performs about 1.5 better labeled F-score on some preliminary runs. fastModel is trained only on clinical data but is a much smaller model and runs a lot faster.
- Introduction to Pipelines
- Piper Files
- Running Pipelines
- Creating Pipelines
- Pipeline Components
- Creating Components
- Piper File Submitter
- UMLS Package Fetcher
- Dictionary Creator
- Simple Pipeline Fabricator
- Pipeline Installation Facility
- Module Basics
- Essential Modules
- for Basic Pipeline
- for Default Pipeline
- for Relations
- for Temporal Information
- for Coreferences
- Creating Modules
- Modules List