<img height="1" width="1" style="display:none" src="https://www.facebook.com/tr?id=2728387060522524&amp;ev=PageView&amp;noscript=1">
Skip to content
Contact us Customer support portal
  • There are no suggestions because the search field is empty.

Confidence Thresholds

In Parashift, the validation steps Separation, Classification, and Extraction each have their own thresholds. These thresholds determine which values are ignored, extracted but not automatically accepted or extracted and automatically accepted.

Confidence Thresholds

Confidence thresholds are defined for each validation step. Generally, these thresholds are set to optimize the dark processing of documents and workflow steps. They can be configured for the workflow steps Separation, Classification, and Extraction.

Separation

The thresholds for separation can be configured in the Upload Profiles and adjusted at any time.
The Separation Threshold is only required if you want to process documents using Intelligent Page Separation. Learn more: Upload Profiles and Intelligent Page Separation

  1. Lower Threshold
    The lower threshold defines the confidence level from which a separation is displayed, but still requires manual confirmation by the validator in the Separation Validation step. 
    1. Red Zone: If the lower threshold isn't reached no separation prediction will be displayed.
  2. Upper Threshold
    The upper threshold defines the confidence level from which separations are displayed and automatically marked as correct. These separations do not require manual validation.

If all separations in a document reach this threshold, the document will be automatically moved to next processing step. 

 

Classification

Classification groups similar documents under the same Document Type and assigns the type that best matches the uploaded document.

  1. The classification threshold defines the confidence level required for the system to automatically assign a document type without manual validation.
    1. Yellow Zone: In this range, the document will require manual classification.

Worth-to-mention: Fallback Document Type
A fallback document type can be set for the classification. If this is active, all documents that do not meet the configured threshold will automatically be classified to the fallback document type. To set the fallback document type, insert the document type ID. Learn more: Upload Profiles

Extraction

For Extraction, thresholds are managed on a field-by-field basis. Each field has individual threshold settings that can be adjusted in the Document Type Configuration.

Two different thresholds can be configured for extraction:

  • Prediction Thresholds define how confident the system is that the correct value has been selected from the document.

  • Recognition Thresholds define how confident the OCR-Engine is that the extracted token has been read correctly.

Learn more:  Prediction vs Recognition

Extraction Confidence Thresholds:

  1. Lower Threshold: 
    The lower threshold defines the confidence level at which a predicted value is displayed but still requires manual verification by a user.
    1. Red Zone: In this range, the prediction will be completely ignored and no result will be shown in the field.
  2. Upper Threshold:
    The upper threshold defines the confidence level at which a predicted value is displayed and automatically accepted as correct. If all fields in a document reach their upper threshold, the document will be dark-processed without any manual validation.

Recognition Confidence Threshold: