Implement Intelligent Document Processing in Forms

Intelligent document processing applies artificial intelligence (AI) and machine learning techniques to capture, extract, and process structured, semi-structured, and unstructured data from a variety of document formats. In Process Automation forms, intelligent document processsing is implemented with the Document Understanding control that uses out of the box pretrained AI models from Oracle Cloud Infrastructure (OCI) Document Understanding AI service to automatically detect, classify and extract texts and objects from uploaded documents.

To know more about OCI Document Understanding AI service, see Oracle Cloud Infrastructure Document Understanding documentation.

Process Automation uses two pretrained AI models from OCI Document Understanding AI service: Document Classification and Key Value Extraction.

• The Document Classification AI model is used to classify the document uploaded to the document understanding control. See Document Classification.
• The Key Value Extraction AI model is used to extract values for predefined key value pairs in the uploaded document. For example, if a passport includes nationality and date of issue, Document Understanding can identify these values and return them as a key value pair. See Key Value Extraction.

Note:

Process Automation is preregistered with OCI Document Understanding AI service for tenancies in the OC1 realm.

To learn how to configure the document understanding control in forms, see Configure Document Understanding Controls.

Configure Document Understanding Controls

Use the Document Understanding control to implement intelligent document processing in forms to automatically detect, classify, and extract texts and objects from files uploaded to it. The extracted text can be used for process routing, approvals or can be sent to a downstream system.

Note:

The document understanding control is available in the Forms palette only if your tenancy is in the OC1 realm. To know more about Oracle Cloud Infrastructure realms, see Regions and Availability Domains in Oracle Cloud Infrastructure Documentation.

To configure a document understanding control:

1. In the Forms palette, expand AI Services and then drag and drop a document understanding control onto the canvas.
2. Select the control and edit its properties on the Properties pane's General tab.

   Field	Description
   Name	An internal identifier that you will use to identify the control.
   Label	The title of the document understanding control that the user will see in the form.
   Accept only these file types	Defines the accepted file types as comma-separated values. If no file types are specified, the control accepts all the files.
   Maximum Upload Size	Defines the maximum file size allowed in megabytes. The default value is 2 MB which is also the maximum upload size allowed.
   Help	Help text that will appear for the control. If you specify help text, a help icon appears next to the name of the control and displays the help text when a user hovers over it or clicks it.
   Primary Text	Text that will appear in the drop-zone of the file picker.
   Secondary Text	Subtext that will appear in the drop-zone of the file picker.
   Disabled	Set this property to display the control as inactive.
   Hide	Set this property to hide the control.
   Events	Allows you to configure events for the control. You can choose one of the following events: On Load: The event is triggered when the document understanding control gets loaded. On Upload: The event is triggered when the file is uploaded successfully. On Error: The event is triggered when file upload or detection/extraction of text in the uploaded file are unsuccessful. You can assign actions or If/Then/Else conditions to the control based on a selected event. See Add Dynamic Behavior to Forms. Note: The Last Uploaded File Name property of the document understanding control allows you to get the name of the most recently uploaded file while configuring events and actions. The Last Uploaded File Content property of the document understanding control allows you to get the content of the most recently uploaded file while configuring events and actions.

3. On the Document tab, edit the following properties.

   Field	Description
   Document type	Specify a document type from the drop-down list. The following document types are supported: Driver license Passport Receipt Invoice
   Create data definition	Select this property to automatically create data definitions for the selected document type. For example, if you chose Passport as the document type then on selecting Create data definition, a passport data object is created. The data object and data definitions are displayed under the Data section. In this section, the document object consists of document and fields sub-objects. The document sub-object is a document reference type and and consists of documentId and documentName attributes. You can use these attributes to reference the document. The fields sub-object contains the related data definitions such as firstName, lastName, country, nationality and so on that are automatically created.
   Generate controls	Select this property to automatically generate controls for the created data definitions. The generated controls are like any other form controls. If required, you can change their position, edit, or delete them.
   Minimum confidence score	Either enter a value, or set a specific value between 0 and 100 using the increment/decrement arrows in the field. Note that higher the confidence score, higher is the chance of the AI model returning accurate results. For example, if you set the minimum confidence score as 96, then the pretrained models must return results that are at least 96 percent accurate.
   Show warnings for low confidence score fields	Select this property to show warning messages for fields with confidence scores that are lower than the minimum threshold.
   Hide fields with no data	Select this property to hide empty fields.
   Hide image preview	Select this property to hide the preview of the uploaded document's image.
   Hide document upload	Select this property to hide the document upload.
   Document display size	Set this property to specify if the image rendered for the uploaded document is Small, Medium or Large.
   Persist document reference in payload	Select this property to include the documentId and documentName details in the payload. These document details are used to reference the document in subsequent user tasks or business flows. Note: This checkbox does not appear for data definition schemas that do not contain the document reference attributes.
   Download document on demand	Select this property if you want the user to click on image to download and display the document. Use this option in scenarios where approvers want to see images on demand, or if there are multiple images and the user wants to avoid longer load times.
   Regenerate data definition	This property appears when you import an application that uses an older schema that does not contain document referencing attributes. Click this link if you want to regenerate the data definition schema so that the new schema includes document reference information. Note: Regenerating is an irreversible action and may introduce breaking changes.
   Messages	Expand this section to view the default messages that will be displayed for the following: Success message Failure message When uploaded document is of different type When uploaded document cannot be recognized When confidence score is too low If required, you can edit the default messages and create messages customized to your business needs.

4. On the Styling tab, edit the control's styling properties.
5. Click Preview to try out using the control.