Question

A financial services firm must prevent the exfiltration of highly structured customer records (account numbers, SSNs) that follow a precise format found in an exported CSV template. Generic pattern-based detection has produced excessive false positives, flagging unrelated numeric strings. The security architect needs a Microsoft Purview classification approach that matches only records that closely correspond to the actual customer database contents while minimizing false positives. Which classification method should be recommended?

Accepted Answer

Configure Exact Data Match (EDM) sensitive information types based on a hashed and uploaded schema of the customer database.

Answer

Use the built-in credit card number and U.S. SSN sensitive information types with a high confidence level.

Answer

Create a custom keyword dictionary containing common account number prefixes and field labels.

Answer

Apply trainable classifiers trained on sample documents containing customer records.