Form Parsing

In this tutorial, we provide example structures of forms and examples of how to parse those.

Form parsing with anchors

If you find a descriptive structure, i.e. anchors, in the form data, you can expect high accuracy extraction results with one document.

Create the following labels

  • Original Approval

  • Current Certificate

  • Certificate Expiry

  • Certificate Number

Form Parsing OCR

Some Labels refer to dates you can change the Data Type to “Date” so that the dates get normalized. The value “31. August 2018” will be normalized to “2018-08-31”.

img_1.png

Form parsing without anchors

In case the form provides no anchors you need to come up with your names of the Labels. As many elements could be parsed we show how to label the Company, Certificate and version of the certificate.

  • Company

  • Certificate

  • Version of Certificate

img_3.png

Multiple lines

Use the lasso to create Annotations which span multiple lines.

Multi-line Form Parser