- Overview
- Model building
- Model evaluation
- Model deployment
- Consuming models
- Consuming models via Flow
- Consuming models via automations
- Consuming models via Document Understanding API
- Developer
- Frequently asked questions
Consuming models via automations
Consume the predictions of IXP models built in Flow Extract nodes in UiPath Studio™, using the Document Understanding activities in Studio Web and Studio Desktop.
In Maestro Flow, the Extract node runs the model as part of your flow, and downstream nodes consume the output of the node.
To consume the model outside of Flow, in a UiPath Studio™ workflow, use the Document Processing activity packages, as described in the following sections.
Prerequisites
- An IXP model built in an Extract node within a solution, deployed to a folder. For details, refer to Deploying models.
- One of the activity packages for the activities that you use, in the following versions or newer:
- UiPath.DocumentUnderstanding.Activities 3.6.0-preview - for cross-platform and Windows RPA workflows.
- UiPath.IntelligentOCR.Activities 7.6.0-preview - for Windows RPA workflows.
Choose one of the two packages, they are not compatible to be used within the same workflow.
Consuming models using the UiPath.DocumentUnderstanding.Activities package
Use the Extract Document Data activity in an RPA workflow to extract data with the model of an Extract node. Optionally, add a Create Validation Task activity to send the extraction results for human review in Action Center.
The variable names used in the following sections are examples. You can name your variables according to your own conventions.
To extract data with the model of an Extract node, proceed as follows:
-
Open or create an RPA workflow in a project/solution.
-
Add the Extract Document Data activity to your workflow.
-
In the Input field, add the file to process. For example, select Create variable, and create a variable of type File. In this example, the variable is called doc.
-
If you are editing a project not part of a solution, or if you do not want to use solution resources (you intentionally switch the Use Solution Resources flag to off), then: a. From the Project dropdown list, select the IXP model of your Extract node. The models of Extract nodes are listed by the name of the node, for example, Extract document (Preview).
b. Select the Version and the Document type of the model.
- Otherwise, if you are editing a project part of a solution, and you choose to use Solution Resources (you leave the flag set to ON). Then, from the IXP Model dropdown list, select the IXP model of your Extract node.
- In the Output section, check the Document data variable, which stores the extraction results. UiPath Studio creates this variable automatically.
For details on all the fields of the activity, check Extract Document Data.
If the model is in a different organization or tenant, set the "Use Solution Resources" check to OFF and configure the following fields in the External connection section, and then select Connect to tenant:
- Credentials Asset - The Orchestrator credentials asset path, in the format
<orchestratorFolder>/<assetName>. - Tenant URL - The URL of the tenant, in the format
https://<base_url>/<organization>/<tenant>.
To use the extraction results in the next activities of the workflow, use the output variable of the Extract Document Data activity from the list of variables, and navigate to the information of interest as per the activity documentation.
Consuming models using the UiPath.IntelligentOCR.Activities package
In Windows projects in Studio Desktop, you can use the Document Understanding Project Extractor activity of the IntelligentOCR package, within a Data Extraction Scope activity.
Defining the taxonomy in Studio
The IntelligentOCR document processing framework requires a taxonomy to be defined in Taxonomy Manager.
To define the taxonomy, proceed as follows:
-
Open Taxonomy Manager from the Design tab.
-
Define the taxonomy in one of the following ways:
-
Import the taxonomy - In Taxonomy Manager, select the Import from IDP projects icon, and then select the published Extract node of your flow. Select the available version and the target document type, select a name you wish to use for that document type in your automation taxonomy, then hit Import.
-
Create the taxonomy manually - Set up your field groups and tables:
- Create a field group or table field that you want to capture data from the Extract node in.
- Add child fields or columns in the respective parent field for each field you need in your automation and is available in your model.
-
-
Add a Load Taxonomy activity to store the taxonomy in a variable, to reference it in the rest of the automation. In this example, the variable is called taxo.
Extracting the data
To consume the model, proceed as follows:
-
Add a Data Extraction Scope activity to your workflow. For the Taxonomy input, add the taxo variable.
-
Within the Data Extraction Scope, add the Document Understanding Project Extractor activity. The Get Capabilities configuration window opens.
-
If the workflow is part of a solution, the Use Solution Resource should be enabled, and then select your model.
-
If the workflow is not part of a solution, or your model is not in the same tenant, select your model from the Project dropdown list. Then, select the version, the tag, and select Get Capabilities.
-
Select Configure Extractors, and use the wizard to map your taxonomy fields to the fields defined in the taxonomy of the Extract node.
Validating the extracted data
Optionally, route the extraction results to human validation. Use one of the following activities, depending on the activity package of choice and validation task type:
- Present Validation Station - attended robot validation task.
- Create Validation Task and Wait for Validation Task and Resume, or Create Validation Task and Wait - Document Validation task in Action Center, after the Extract Document Data activity.
- Create Document Validation Action and Wait For Document Validation Action And Resume - Document Validation task in Action Center, after the Data Extraction Scope activity.
- Create Document Validation Artifacts and Create App Task - Low-code or coded app tasks in Action Center that use Document Validation components, paired with their counterparts, Wait for App Task and Retrieve Document Validation Artifacts.
For details on validation, check the following resources: