Extract Document Values with manual validation

The Extraction Validation user interface is a central component of the Parashift Platform App. You use it to manually validate automatically extracted data and to train new fields. It’s designed to be easy to use for casual users while also catering to power users. This article provides insights into the different components and best practices for using Extraction Validation.

Validation Interface

The interface can be divided into the following five areas:

Breadcrumbs

Displays some document information, such as:

1. The path to the menu from which Extraction Validation was opened (clickable for quick navigation).
2. The document ID and document name, with the ability to change the document name.

Actions

Shows the document type and the available actions for a document.

1. Document Type – Displays the current document type.
2. Back – Returns to the list from which the user entered Extraction Validation or the main Extraction list.
3. Switch between Single and Serial mode – See Serial Mode in-depth explanation.
4. Secondary Actions – See Done, Save, Force-Done, and Forwarding to 1st/2nd/3rd Level.
5. Finish Validation – This button is only active when all fields are properly validated (see Done, Save, Force-Done, and Forwarding to 1st/2nd/3rd Level).

Viewer

Allows you to view the document, navigate through pages, and more:

1. Hide Sidebar
2. Jump to Value Option
3. Rotation – If preprocessing didn’t rotate your document correctly, rotate it manually.
4. Zoom In/Out – Zoom in or out for a closer look.
5. Document Viewer Options and Highlighting Options.  

6. Page Navigation
   Navigate easily through the pages of the document using the Page Selector. This tool enables you to:

   • Jump to the first or last page
   • Skip forward or backward page by page
   • Enter the specific page number you wish to access directly

Editor

 

Displays all extracted data together with the Info-Box/Field section, forms one of the most important parts of Extraction Validation.

Fields are grouped into Sections and Field Sets. The layout, size, and section names are either preconfigured (when using our standard document types) or customized by an administrator. The Field Editor uses lazy loading, ensuring fast and reliable performance even with large amounts of data.

Line Items Extraction with Table Input

A Table Input feature view has been created for the validation line item section to provide a more efficient and user-friendly experience. More information here: Table Input

Info-Box

Document

Provides general information about the document, such as the document owner, document type, recognized language, average recognition confidence, and relevant upload parameters and dates.

Hover over any title or data point for a more in-depth description.

 

Field

 

This tab opens by default when navigating through fields. Together with the Field Editor, it’s one of the most important screens in Extraction Validation.

 

1. A preview of the selected value on the document (useful for quickly comparing the read value to the actual text in case of low OCR confidence).
2. Field status and any open error or warning messages.
3. Prediction Confidence – How sure the machine is that it predicted the correct value. If a user manually picked the value, a user icon appears.
4. Recognition Confidence – How sure the machine is that it read the value correctly (OCR/Barcode, etc.). If a user manually corrected the value, a user icon appears
   

Learn more: Difference between Recognition and Prediction

Tags & Flags

Change whether the document should be part of the general training pool, mark a document as unprocessable or add a tag.

Comments

Leave a comment on the document or read comments from other users. If a document has comments, a small notification bubble appears on this tab.

Field Validation – Status, Warnings & Errors

A field generally has three different states

1. Valid (Green or Deactivated Look)

   • Fields are valid if configured thresholds are met (e.g., predicted with high confidence), the field is optional, or it’s already validated. They appear in gray to indicate they don’t need attention.

2. Warnings (Yellow)

   • Fields with warnings need user action or confirmation.

3. Errors (Red)

   • Fields with errors always require manual user interaction and cannot be confirmed without resolving the issue.

Field Valid

The platform distinguishes between automatically valid fields "Field valid."  (no user interaction) and valid fields that required user interaction "Validated."

Warning

A field can have one or more warnings. Each warning must either be corrected (value changed) or confirmed (by pressing ENTER).
Most Common Warnings

Error

A field can have one or more errors, each of which must be resolved by correcting the value. 

Most Common Errors

Navigation

General

The jump order goes from top-right to bottom-left.

On Opening

When opening a document, focus is on the first invalid field; valid fields are skipped.

Mouse Navigation (For Casual Users)

You can click any field in the Editor to edit it.

Keyboard Navigation (For Power Users)

Press ENTER to move to the next invalid field, skipping valid fields so you can focus on fields that need attention. If there are no more invalid fields, you can press ENTER again to finish Validation.

Although you can also navigate with TAB, it does not consider the field state.

End of Document

When you reach the end of the document:

• If any warnings/errors remain, focus returns to the first invalid field.
• If no invalid fields remain, a pop-up indicates that validation is complete. Press ENTER again to end Validation. Depending on Single or Serial mode, the next document either opens immediately or the list view loads.