Sections
Sections are the parts of a document that you want to leverage in the symbolic rules of an explainable model.
For example, in an email message categorization model, it can be better to give higher importance to the subject. Another example: in a model for extracting information from contracts, better to look for the covenants only in the section of the contract dedicated to them.
Explainable models generated with the Platform authoring application contain symbolic rules that do not take sections into account, but Studio allows you to edit those models and refine the rules to give them a section scope.
At the end of the development cycle, the model ends up as a block of an NL Flow workflows: how can the block identify the sections in the text submitted to it? Well, it doesn't have to: it's up to the user of the model to specify in the input JSON where each section begins and ends, that is the section boundaries.
Studio and Platform are integrated, and models generated in Studio can be easily imported into a Platform authoring project to be tested and determine their quality, also in comparisons with other models of the same project.
If the Studio-generated model leverages sections, all the documents of the test library should have information about section boundaries. This is achieved by annotating sections: for each document, users select a portion of the text and choose the corresponding section from a list. This in turn implies that possible sections must have been previously defined as project resources.
Follow the links below to find out how to define sections at project resources and annotate their occurrences in document text.
-
Definition:
When uploading documents from an exchange archive, there may be annotation files in the archive, and between the annotations there may be section annotations. The upload wizard has options to manage the harmonization between the sections annotated in the archive documents and those defined at project level.