Data Deanonymizer
Data Deanonymizer does the reverse of Data Anonymizer: it takes a text in which personal data was replaced with dummy values and puts the real data back.
To do this it uses the mapping between real and dummy values that Data Anonymizer saves in each record. It is the final step when you have worked with AI on anonymised information and need the result with the real data, for example to reply to a customer.
When to use it
- Restore the customer’s name and contact details in a reply an agent drafted from an anonymised text.
- Produce a final report with the real data after analysing it anonymously.
- Return documents that were processed without personal data to their original form.
Configuration
| Field | What it does |
|---|---|
| Data | The records with the text you want to restore. Each record must also include the data mapping saved by Data Anonymizer. |
| Source data input key to deanonymize | Name of the field holding the anonymised text. Default: text. |
| Deanonymize Mapping Key | Name of the record field that holds the mapping between real and dummy values. Default: anonymizer_mapping, the name Data Anonymizer uses. |
| Deanonymize Mapping | Optional. Connect the output of Data Anonymizer here when the text to restore is in a different record (for example, an agent’s reply): its mapping, under Deanonymize Mapping Key, is used for every record. Unconnected, each record’s own mapping is used. |
Output
The Data output returns the records with the text of the chosen field and the dummy values replaced again with the real ones. Replacement ignores upper and lower case.
Example
An agent drafts replies to complaints without seeing customer data.
- Data Anonymizer replaces the personal data in each complaint.
- The agent writes the reply on the anonymised text.
- Store the reply in a field of the same record, next to anonymizer_mapping.
- Data Deanonymizer, with that field in Source data input key to deanonymize, returns the reply with the customer’s real name, ready to send.
Good practices
- Keep the data mapping in the same record as the text from the moment you anonymise it until you restore it.
- If you enabled Remark Anonymization in Data Anonymizer, the asterisks around replaced values stay in the text.