document-understanding

2021.10

false

OUT OF SUPPORT

Document Understanding User Guide

DELIVERY:

Automation Cloud Automation Cloud Public Sector Automation Suite Standalone

Last updated Nov 11, 2024

Validation Station

Important: In case you use an Intel Xe GPU and Validation Station is not displayed properly, we recommend updating the graphics driver to the latest version. For more information, please visit this page.

The Validation Station is a user interface that can be used to validate and correct data extracted from documents using the Document Understanding framework.

Validation Station Management

Validation Station can be used both as a stand-alone tool and integrated with Orchestrator/Automation Cloud Orchestrator.

Also, it can be used either as an attended activity with the Present Validation Station or as a type of action in Action Center with the Create Document Validation Action, Wait for Document Validation Action and Resume, as part of Orchestrator Tasks.

Requirements for Document Validation

Certain inputs need to be provided for the Validation Station to function. Here is a list with the minimum required inputs in order to configure the Validation Station as an attended activity:

Provide a Taxonomy needed for processing the document. It can be created directly in the workflow by using the Taxonomy Manager wizard, and it can be loaded using the Load Taxonomy activity. It should be provided as a DocumentTaxonomy variable.
Provide the Document Path for the file you want to validate.
Provide a Document Object Model for the file, which can be obtained as an output of the Digitize Document activity.
Provide the Document Text of the file, which can be obtained as an output of the same Digitize Document activity.
Provide the Automatic Extraction Results which represents the data automatically extracted by the robot, that now needs human validation. This is the output of the Data Extraction Scope activity.

Note: If no variable is provided in the AutomaticExtractionResults property field, or if no classification is published in the provided variable, the Validation Station opens in full manual processing mode.

Validation Station as an Attended Activity

You can use the Present Validation Station activity anywhere in your workflows.

An attended activity is functional only if the file that needs to be processed is available locally on the machine that executes the process.

At run-time, a window opens and displays the Validation Station allowing any interaction from the user side.

Validation Station in Action Center

Increase your productivity by creating an orchestration process that adds document validation actions in Action Center, in both On-Prem Orchestrator or its Automation Cloud Orchestrator counterpart. This action reduces your need for storing the documents locally, having a robot installed on each human's operated machine, or having the robot wait for human users to finish validation.

This approach requires you to have your robot connected to your Orchestrator and to build a project of type Orchestration Process in UiPath Studio. You can publish your project to Orchestrator once it is ready. Here you can find more details about this process.

When running a project with UiPath Orchestrator, you need to make use of the Create Document Validation Action and Wait for Document Validation Action and Resume activities because this is how you are creating the connection between the unattended robots executing the automatic part and the Knowledge Workers that can pick up actions from Action Center (on-premises or cloud-based).

In this scenario, the Validation Station UI is displayed directly in UiPath Orchestrator's Action center but has the same functionalities like the one run as an attended activity.

Using Validation Station

The Validation Station is a powerful tool that enables users to review and correct automatically extracted data from files, or manually process files for data extraction.

The Validation Station, once opened, present all extracted information along with the file being processed:

The fields that are visible in the Validation Station are the ones defined in the Taxonomy used in your workflow.

Document View Section

The right area of the Validation Station contains an interactive version of the original document, in which text or document sections can be selected, and words can be clicked based on the output of the digitization process.

This area also contains options for zooming in and out, selecting and rotating pages, searching through the document, or switching to text view.

Here are the options available in the right part of the Validation Station screen:

Option	Description
	Toggles between the text view and image view of the document.
	Zooms in on the document.
	Resets the zoom level on the document. This option is enabled only if the document was previously zoomed in or out.
	Zooms out on the document.
	Rotates the current page clockwise.
- Tokens - Custom area - Choice on selection	Sets the selection mode while in image view
- Text - Tokens	Sets the selection mode while in text view
	Initiates a search between results in the document used by the Validation Station.
	Displays all the available keyboard shortcuts supported by the Validation Station.

Here is an example of how to use the Toggle, Zoom In, Zoom Out, and Reset Zoom options:

To zoom in or out, you can also use the CTRL+scroll mouse wheel combination: CTRL+scroll up to view a specific section of the document; CTRL+scroll down to view a larger section of the document.

The below example illustrates the Rotate option:

Area selection in image view can be observed below:

Correspondingly, please look at area selection in text view:

The next example explains how to use the Search option:

There are many keyboard shortcuts that can be used to optimize the human interaction with the Validation Station. We encourage you to use them as much as possible. These can be reviewed in the Keyboard Shortcuts popup.

Shortcut	Description
n	Moves to the next field
p	Moves to the previous field
f v	Marks a value as validated
f c	Changes the extracted value
f z	Reverts to the previous value
f a	Adds an additional value
f s	Toggles between suggestions
ESC	Exits edit mode (for Fields and Tables) Collapses the derived parts (for Fields) Deselects a line or Exits table selection mode (for Table Selection)
DEL	Removes the selected value (for Fields) Removes the selected line (for Table Selection)
d +	Zooms in
d -	Zooms out
d 0	Resets zoom
d r	Rotates the page clockwise
d t	Toggles the text mode
/	Initiates a search
d s	Changes selection mode
→	Moves to the right cell
←	Moves to the left cell
↑	Moves to the top cell
↓	Moves to the bottom cell
t v	Marks a cell as validated
t c	Changes the extracted cell
t z	Reverts to the previous cell value
t DEL	Removes the selected cell
s ←	Moves the selected line to the left
s →	Moves the selected line to the right
s ↑	Moves the selected line up
s ↓	Moves the selected line down
s d	Duplicates the selected line
s v	Selects vertical lines
s f	Selects horizontal lines
s a	Auto-detects lines by mouse movement
s t	Hand tool selection - moves and deletes lines
?	Opens the keyboard shortcuts list
!	Reports the document as an exception
CTRL Enter	Saves all data
CTRL DEL	Discards all current changes
1 2 3 4 5 6 7 8 9 q w e r y u i o a g h j k l b m @ # $ % ^ & *** ( ) [ ] {	Assigns values to a field Selects the corresponding field

Visualize the available keyboard shortcuts:

Data Extraction Section

The left area displays the document type you have selected for the current validation and enables you to select the state of each element and link it to its corresponding word or area in the document.

You can use the field shortcuts to assign values to a field or to toggle between fields. Once a value is assigned to a field, it is highlighted in accordance with the color of the selected field.

For the assigned value, there is a document crop displayed in the table field. This helps with better locating the area from which the value was extracted and it also serves as a means of double-checking the value by comparing it with the document crop.

Note:

The Document Type field is a special field that you can act upon:

If the extraction results contain a document type, and that document type is correct, then no action is required.
If the extraction results contain a document type, and that document type is incorrect, then you have to select the correct one and provide evidence for it from within the document.
If no extraction result is provided and only one document type exists in the taxonomy, then that document type is pre-selected but needs evidencing.
If no extraction result is provided and there are multiple document types in the taxonomy, then you have to manually select the desired document type and provide evidence for it.

All fields that contain information have an Options drop-down menu that can be accessed by clicking on the button. A drop-down list becomes visible, displaying multiple editing options.

The menu includes the following options:

Change extracted value - Changes the automatically extracted value with a manually selected one. This field is active only when one or multiple values are selected from the document and are different from the original value.
Revert to previous value - Resets the field's value to its last state. This option is active only when a value was previously altered or deleted.
Mark as missing - Marks a field as missing if the information is not available in the document.

Selection Modes

There are several ways of selecting text while using the Validation Station wizard. Using them allows you to quickly navigate through the entire document and easily select the desired words for validating a field.

Here is a list of all the available selection options:

Select one word - Click the desired word.
Select consecutive words - Click the first word, then SHIFT+click the last word from range.
Select multiple disparate words - Click the first word, then CTRL+click the rest of the desired words.
Combine multiple selections - Click the first word, then SHIFT+click the last word from a range for the first selection, then hold CTRL+click and SHIFT+click to add another range, until you've completed your selections.
Area selection - Make a selection and choose the selection type:
- Tokens - Selects all words in the selected area.
- Custom area - Captures only the area and not the words in it.

Other Options

Edit the field's value - Changes the content of a field by positioning the mouse on that field, clicking the value, and adding the desired input.

The option - Reverts the field to its prior state. One-click on this option takes you one step back, meaning that if you had several changes on that field, multiple clicks might be required for returning to a certain value. This field is active only when a value was previously modified or deleted.

The option - Adds a value to the field by using the Custom area or Tokens selection. The button becomes available when a selection is made in the document and differs from the one in the field. The selection can be made for multi-value fields at all times, and for single-value fields only if no value is present for that field.

The option - Confirms the information included into the field. Once confirmed, the check box turns blue, like this .

The interface of the Validation Station is interactive, meaning that when a field is selected on the left side, the right side moves the focus on it by highlighting it.

This also applies to text view:

The option - Enables you to select and add additional values from the document to a specific field.

The option - Enables you to add a value to a field without requiring reference from the document.

Table Fields - Cell Level Processing

Below you can find explanations for all the options available for a table field:

Icon	Description
	Indicates the table's area in the document.
	Opens a drop-down list with the following options: Revert to previous value - Returns to the previous value. It is active only if the table's data has been previously altered. Remove value - Marks the table field as missing.
	When selected, confirms the data accuracy.
	Enabled when the field's value is missing.
	Enabled when the initial extracted value had been altered or deleted. When selected, it restores the previous value.
	Removes the value, in this case, the existing table.

All table fields have a drop-down menu with the following options:

Change extracted value - Selects a new value from the document and replaces the initial one. This field becomes available only when the newly selected value is different from the original selection.
Revert to previous value - Replaces the actual value with the previous one. This field is active only when altering the initial value.
Mark as missing - Marks a field as missing if the info is not available in the document.

The drop-down menu available at the end of each row has the following options:

Make it the table header - Transforms the selected row in the table's header. The row cannot be seen in the table's body anymore but the information can be visualized any time a header's field is selected.
Insert row above - Allows you to add a new row, above the selected one. The row is automatically added to the table, except that all fields are marked as Not extracted. You have to manually select the value from the document and add it by using the Add extracted value option. An example of how to use this option is illustrated below:
Insert row below - Applies the same principle as for the Insert row above option, the only difference being that the row is inserted below and not above the actual selection.
Delete row - Deletes the selected row.

You can define the table header when using the Extract new table option by enabling the Extract header function, by selecting the information from the document, or by transforming one of the existing rows into the table's header.

Extract the table with a defined header by enabling the Extract header function once the Extract new table option is active.
Confirm the table's header by positioning yourself on the header and selecting its equivalent from the table. Once the value is selected, click the Change extracted value from the drop-down menu.
Transform existing row into table's header by positioning yourself on one of the rows and selecting from the drop-down menu the Make it the table header option.

Table Fields - Table Level Processing

A table can manually be selected and defined, straight from the Validation Station wizard. If no table is selected, or if you are not happy with the automatic selection, then you can use the options available on the drop-down menu found at the end of the first row.

Note:

If the Value of a table cell is not extracted, you can manually add a value into that table cell, by going into the Selection Mode, choosing the Custom area option, and marking the table cell area.

Both Extract new table and Extract Rows from here options are using the same functionality and are enabling you to define new values.

Extract new table - Replaces the existing table with the new selection. You need to mark all rows and columns. Please keep in mind that the first row becomes the header of the table.
Extract rows from here - Enables the custom selection tool and offers you the possibility to manually select new rows and add them below the currently selected one. Once the area of the new rows is selected, you must define each column. You can do this by using the available options presented in the table below.
Highlight row - Highlights in the table the selected row.

Highlight table - Highlights the entire extracted table area.

Delete all rows - Deletes the existing table.

Icon	Function
	Removes all lines visible in the selection.
	Removes only the selected line from the selection.
	Enables horizontal lines in the selection.
	Enables vertical lines in the selection.
	Enables you to adjust the line's direction using the mouse.
	Enables the selection, rearrangement, and removal of lines.

Below you can see an example of how to extract a new table row and how to use the available options. The Extract header function is not available when appending rows.

Here is how you can extract a new table by using the Extract new table option:

You can click the Save new table button to automatically confirm all fields or you can deny the operation by clicking the No button, return to the table selection, and manually confirm each field.

Note: If you want to save a table with empty or no extracted fields, then those fields are automatically marked as missing.

Value Formatting and Language Setting

Number, date, and address fields allow you to review and correct formatted (parsed) parts of a specific value. The editable parts are as follows:

Field Type	Editable Formatted Parts
Number	Value
Date	Day Month Year
Address	Address Line 1 Address Line 2 Address Line 3 City State / County / Province Country Zip Postal Code
Name	First Name Middle Name Last Name

When you extract or correct a value for a field of any of these types, the Validation Station tries to automatically parse the value into its formatted components.

The language setting displays the detected prevalent language within the document, as identified during the digitization process. This enables the Validation Station to parse numbers and dates more accurately, according to the language of the document. You can change the language setting by using the drop-down menu.

By doing so, when you manually extract or change a date or number value, the Validation Station will first try to format the selected string according to the selected language, and will fallback to English US if parsing is not successful as such. The formatting function only applies to the editable formatted parts of a value, not to the original string value.

To ensure the best automatic formatted value detection, we recommend you check the detected language and correct it if necessary.

Report Exceptions

You have the option to report a document as an exception. If this situation occurs, the Present validation Station throws an exception that should be caught by the RPA workflow and treated separately. The exception message displays the Reason for Exception filled in by the user.

Here is how you can report a document as an exception:

Data Confirmation & Validation

You have the option to manually or automatically confirm all fields. For manual confirmation, you need to select the check box of each field. If a check box is not manually confirmed, then this process is automatically done when the Save button is clicked, and then the action is confirmed by clicking the Continue & save button.

The following options are available on the bottom side of the Validation Station:

Button	Function
	Saves the confirmed fields.
	Saves and closes the table selection area. The button is enabled only when the table field is active.
	Reports the document as being an exception.
	Enabled only when not all values are manually confirmed. By selecting it, all data is automatically confirmed and saved.
	Enabled when no change has been done to the table.
	Enabled after a change has been done to the table.
	Enabled after the user clicks Dismiss. Discards all changes done to the table.