Skip to main content
Skip table of contents

Form

This function automatically extracts key values contained in documents.

The Intelligent Document Data extraction employs advanced Artificial Intelligence (AI) technology to streamline the extraction of data from documents, especially structured documents (forms), making it a powerful tool for enhancing efficiency and productivity.
It is important, however, to acknowledge the limitations of the AI in certain scenarios to manage expectations effectively. The success of the extraction process relies on the AI's ability to establish clear relationships between keys and values in the form. In instances where this relationship is unclear or ambiguous, the AI may struggle to provide accurate results.
Users should be mindful that the AI is a supportive tool rather than an infallible solution, and that not all forms may be processed with equal precision. The extraction can differ depending on the selected capture method, the scan settings, or other factors.
To ensure the best results, the extracted data should be reviewed for accuracy using the RSI LogicFlow validation feature, especially in cases where form layouts are unconventional or key/value relationships are ambiguous, for example when the value is not in the immediate vicinity of the key.
It is recommended to always enforce the RSI LogicFlow validation when the extracted data accuracy is critical.

To configure the Intelligent Document Data, the expected keyword should be entered. This is the information the AI will search for to extract relevant data, so it is important to match the keyword as it is written on the document. For each extracted data, it is possible to look for various keywords in order to support different document formats. The AI will select the best value found on the document.
It is possible to narrow down the data extraction to specific content within the document with the content filter and the page filter, ensuring you get the right data even if the key appears multiple times.
The data type should be selected and a unique Data ID should be assigned to the extracted information, serving as an identifier for subsequent workflow use and as a label in the validation interface.


Details of settings:

Category

Setting

Description

Options

Data to be Extracted

Search Label

Label to search in the document.

Any string.

Content filter

Allows filtering key-value pairs.

Can filter using:

  • Off: no filtering

  • Beginning with: only value starting with the specified string.

  • Ending with: only value ending with the specified string.

  • Containing: only value containing the specified string.

  • Not containing: only value not containing the specified string.

  • Match regular expression: only value matching the specified string.

Page filter

To specify a specific page to analyze

Must be a number.

Data Type

To specify the data type.

The type can be:

  • Text

  • Number

  • On/Off (Checkbox)

  • Date

Data ID

ID for using the extracted data in the input fields of other functions. You can leave it as the default or change it to an original ID that is easier to identify.

Any string except “%”.

Input Method

Choose how to define the list of acceptable values for this field.

Only available to configure on Text or Number type fields. For other types, Use extracted value directly will be used.

Select from List: Define a list of predefined options. The AI will attempt to match the extracted value to the closest option in the list. Users can also manually select from the list.

Use extracted value directly: The AI extracts the value directly without matching against a list. Users can provide values freely.

CSV file for lookup (Select from List only)

Import multiple values at once from a CSV file. The CSV file should contain one value per line (or comma-separated values on a single line).

Note: Importing a CSV will overwrite all existing values in the list. Supported format: .csv files.

Supported format: .csv files with values separated by commas or line breaks.

List values entry

Add individual values to the list one at a time. Values are validated based on the field's data type (text or number).

Text fields: Enter any text value. Duplicate values are not allowed.

Number fields: Enter values in US format (1,234.56) or European format (1.234,56).

Values can be reordered using ↑↓ arrows or removed with the × button.

Allow users to manually enter text not listed in the options.

Controls whether users can enter custom values not in the predefined list.

Checked (Open List): Users can select from the list OR type their own custom value in a text input field. As the user types, the list dynamically filters to show only matching options.

  • Custom values entered by users that are not present in the list are marked with a yellow indicator.

Unchecked (Closed List): Users can ONLY select from the predefined values in a strict dropdown menu. No custom input is allowed.

  • If the AI extracts a value not in the closed list, it is marked with a red indicator, and with an AI Added suffix displayed.

JavaScript errors detected

Please note, these errors can depend on your browser setup.

If this problem persists, please contact our support.