Setting Up Anonymization (PII) and Custom Recognizer

Masking Personal Identifiable Information (PII) for Data Source Fields

To configure masking of PII for data source fields:

1. Navigate to the Settings > Data Sources menu of the Aisera Administration application to specify your data source configuration.

   

2. Select the embedded application for which you want to mask fields in the data source.

   

   The PII configuration is set at the field level for each data source.

3. Hover your cursor over the lower-right section of the window to see the edit button for the field you want to configure.

The following screenshot shows the Description field with masking applied to it for a few standard entities.

Here’s a template you can use that will identify and scrub the following out-of-the-box entities:

• PERSON

• PHONE

• EMAIL

• CREDIT CARD

The fields that need to be masked should be included in the AnalyzeTemplate object and the value it will be masked with should be added to the AnonymizeTemplate object.

{ "AnalyzeTemplate": { "fields": [ { "name": "PERSON" }, { "name": "PHONE_NUMBER" }, { "name": "EMAIL_ADDRESS" }, { "name": "CREDIT_CARD" } ] }, "AnonymizeTemplate": { "fieldTypeTransformations": [ { "fields": [ { "name": "PERSON" } ], "transformation": { "replaceValue": { "newValue": "<PERSON>" } } }, { "fields": [ { "name": "PHONE_NUMBER" } ], "transformation": { "replaceValue": { "newValue": "<PHONE_NUMBER>" } } }, { "fields": [ { "name": "EMAIL_ADDRESS" } ], "transformation": { "replaceValue": { "newValue": "<EMAIL_ADDRESS>" } } }, { "fields": [ { "name": "CREDIT_CARD" } ], "transformation": { "replaceValue": { "newValue": "<CREDIT_CARD>" } } } ] } }

For entities that are not available as out-of-the-box entities, you will need to create a custom recognizer and specify the exact pattern of the data.

For example: Server id can consist of more than 12 alphanumeric characters and some special characters or has the format ID-12345678. So you have to provide the pattern for each of these entities.

If a pattern is too generic, such as 8 numbers, then we cannot create a custom recognizer because it will have a lot of false positives.