Establishing document types

To create the document types for your classification model, you upload your sample documents to Document Processing Designer to be examined and categorized.

Procedure

To create a document classification model:

  1. From the Automation in Business Automation Studio, click Open on the project tile or click the project tile then click Open for a specific version.
  2. In the Document types and samples tile, click Start.
    The Document types and samples screen includes a list of pre-trained document types that are included with the default model.
  3. Click Create document types, and choose how your sample files are organized:
    Sample documents Steps
    With my sorted samples Choose this method if you have already determined which of your documents belong to the same type.
    1. Enter a value for the Document type display name. You can set this field in any language, using any Unicode character.
    2. Enter a value for the Document type symbolic name. This name is used to reference this project in the code, and cannot be changed. It cannot have spaces or special characters.
    3. Enter a classification confidence threshold. The confidence level is shown in the training results of your classification model, expressed in percentage. If the confidence level is under the threshold that you set here, there will be a warning in the document. The default value is 70%.
    4. Upload your samples.
    5. If your documents have a fixed structure, such as tax forms, select Fixed format.
    6. If you want to include feedback documents in your training, select Feedback documents. Then, you can adjust the percentage of corrected documents you want to use in retraining.
    7. Click Done to add the document type with the samples that you uploaded.

      You see the document type and a list of samples. You can view a sample document by clicking the document name.

      If you selected fixed-format document type, you are able to upload empty samples, which help the model distinguish document characteristics from data. If you edit the document type and clear the fixed-format option, empty samples are no longer available.

    8. To add another type, click Create document type and repeat these steps.

      You can also select single or multiple documents from the samples list, and use the controls on the banner to assign the samples to a document type or to delete the samples.

    9. When you have added all of your document types, click Finish.
    With my unsorted samples Choose this method if you have not sorted your sample documents into types.
    1. Click Upload to add your sample documents to the project.
      Remember: To get enough information into the model, you should provide at least five different samples for each document type that you want to include and at least one document for testing. Make sure that you have enough samples of the types you are modeling to use when you train the model.
    2. Click Categorize.

      The system sorts your samples into categories.

    3. Click a category, then click a document in the list to view or check which documents were grouped together into that category.
    4. On each category, click the Overflow menu Three dots menu and choose one of the following options:
      • Assign to new document type
      • Assign to document type
      • Delete

      You can also select single or multiple documents from the samples list, and use the controls on the banner to assign the samples to a document type or to delete the samples.

    5. When you have created a document type from each relevant category, click Next.
    6. Review the document types that you have created from the categories. You can click any document in the list to confirm the correct categorization for the document. When you are satisfied with your document types, click Finish.
    Without samples You have the option to create document types without providing samples to the system. The Document types page displays the list of pre-trained document types that are available in the project by default.
    1. Click a document type to see the fields that are available.
    2. Add or edit fields as needed.
    3. Optionally create a new document type by clicking Create document type.
    4. When you are satisfied with your document types, click Finish.

Results

The focus returns to the main Designer page.

What to do next

Use the Overflow menu Three dots menu as needed to edit or delete your document types.