Data classification concepts

Classification involves identifying and labeling content in your organization to better understand your data landscape.

Sensitive information types

Sensitive information types allow you to detect and label common types of sensitive data automatically. This helps ensure sensitive data is automatically identified and protected, reducing the risk of data breaches. Microsoft Purview provides over 300 built-in patterns, such as credit card numbers, Social Security Numbers, and other regulated data. These patterns are identified through regular expressions or functions. For organizations with unique needs, you can create custom sensitive information types to capture proprietary or specialized data.

Trainable classifiers

Trainable classifiers use artificial intelligence and machine learning to identify content specific to your organization, such as contracts or customer records, without relying solely on pattern matching. These classifiers reduce the risk of missing sensitive content that doesn’t fit standard patterns, offering more comprehensive protection.

Exact data match (EDM)

Exact data match (EDM) provides a highly accurate way to identify sensitive information by matching specific, predefined data values from a secure data source. This is especially useful when your organization deals with highly sensitive and structured data, like employee IDs or customer account numbers. EDM helps prevent false positives by precisely targeting specific data, rather than relying on general patterns or keywords.