Classification

Classification is the assignment of a document type to a document during production. If extraction is configured for your solution, successful classification is needed for successful extraction.

DocAI Studio supports three classification types.

Layout classification:

Uses the layout of training documents uploaded during configuration. (Default: Enabled)

Content classification:

Uses the content of training documents uploaded during configuration. (Default: Enabled)

Generative AI classification:

Uses the description provided in each document type. No training documents are required. (Default: Disabled)

Classification sequence

By default, the system performs classification in the following sequence:

  1. Layout Classification

  2. Content Classification

  3. Generative AI Classification

The sequence stops as soon as the system returns a confident classification result. If you disable one of these steps, the system skips that classification step.

With the default settings, the system performs generative AI classification only when both layout and content classification do not return a result or when no training documents exist.

However, you can change the classification sequence when using generative AI classification. Enable the Run before other classifiers setting to run generative AI classification first. The system then runs layout and content classification only if generative AI does not classify a document.

More info