Restoration
Restore the original values of pseudonymized content with a single click.
Restoration is the complementary tool to pseudonymization. It allows you to restore the original values in content that has been previously pseudonymized or generated from pseudonymized documents, most often by an AI assistant.
How It Works
Pseudonymization replaces each identifying piece of data with a consistent pseudonym (a unique label that remains consistent throughout a document or corpus of documents):
| Data | Associated Pseudonyms |
|---|---|
| Marie Dupont, Marie, Ms. D. | PR1a, PR1b, PR1c |
| Thomas M., Thomas Martin, T. Martin | PR2a, PR2b, PR2c |
| Technoplus SAS, TECHNOPLUS | CIE1a, CIE1b |
| Vival Group, VIVAL GROUP, VIVAL | CIE2a, CIE2b, CIE2c |
The technology creates a unique pseudonym for all syntactic variants of an entity, and closes it with a letter naming the variant written at each point of the document, unless the session merges its variants.
Restoration puts back the exact variant a lettered pseudonym names: PR1b becomes "Marie" again, so a document edited by an AI assistant keeps its original wording. A pseudonym without its letter, as an assistant writes it in an analysis (PR1), becomes the reference value of the entity.
It maintains the mapping between each original value and its pseudonym in the session’s pseudonymization table, accessible from the Pseudonyms tab, and uses this table to perform the restoration (de-pseudonymization) operation.
Usage
- Go to the Restoration tab of a session
- Paste the content to be restored (text, AI response, etc.) or import a document into the designated area
- The technology automatically replaces each pseudonym found with its original value
- Download or copy the restored content to share it