Smart Redaction

From manual to smart redaction

Manual redaction

Manual redaction is still commonly used in day-to-day office work.
It implies several steps:

• Searching for the right information to redact
• Manually selecting regions to redact

This solution is time-consuming and prone to errors, as the search phase relies on human eyes only. A fully manual process is also hard to scale.

Search and redact

With the help of an OCR engine, redaction is much faster and more accurate.

The use of regular expressions (RegEx) makes processes much faster. In addition, it is very useful when the information that needs to be redacted is identified beforehand. For instance, we know we want to redact social security and phone numbers.

However, a search process that relies on a traditional OCR only is not performant on scanned documents and often doesn’t retrieve information trapped in tables and graphics.
A manual search will still be needed to verify that all information has been redacted.

Smart redaction

The GdPicture.NET smart redaction engine goes a step above, thanks to natural language technologies and computer vision.

• It speeds up and simplifies the redaction process significantly.
• The engine is not looking for specific text but rather a type of information. As a result, the amount of information redacted at once is much broader.
• The results are automatically redacted; no need to select the regions to redact anymore.
• It works on scanned, low-quality, and skewed documents.
• It redacts data in tables, graphics, colored backgrounds, and text in various orientations.
• It automatically retrieves credit card numbers, email addresses, IBANs, phone numbers, URIs, VAT IDs, VINs, SSNs and more.
• It is built for automation.