How to build an NLP pipeline?

NLP pipelines combine a sequence that may contain:

  • Converters to transform source formats
  • Custom-made or off-the-shelf modelsannotators and processors to manipulate text or annotations,
  • Formatters to provide the desired output format.

To create a pipeline:

  • Go to the Pipelines view
  • Create a new pipeline

  • Give a name to the pipeline.
  • Select and combine components:
    • Document conversion: select a converter

  • Processing pipeline: select a sequence of modelsgazetteersannotatorsprocessors
  • When adding a component to a pipeline, you can select either:
    • Project annotators that are part of existing projects or
    • External annotators independently of any existing project.
  • You can add conditions under which a particular annotator is launched
  • Output format: select a formatter
  • Save the pipeline with the tick button at the top right of the window.
  • Then go to the Annotation Tests view to test the pipeline with the text of a new document. Pipelines are all listed in the drop down menu alongside models, gazetteers, suggestions models of your current project.

See also: