Covers the methods Purview uses to recognise sensitive content, from pattern-based types and exact data match to machine-learning classifiers, fingerprinting and OCR. Labelling, DLP and retention all rely on what it detects.
Read more: Microsoft Learn
In the Ultra Transcenders books
Each book explains Data classification in context, with comparison tables and the common traps.
Terms in this definition
- Exact data match
A custom kind of sensitive information type that looks up the real values held in a table your organisation supplies, such as a customer list, rather than matching a generic pattern. Data loss prevention can then flag only those precise records.
- OCR
Optical character recognition. It reads handwritten or printed text from documents and images, making it available for analysis, something Azure Language cannot do with scanned images by itself.
- Microsoft Purview Data Loss Prevention
Policies in Purview that look for sensitivity labels or sensitive information types in content held in many places, including Microsoft 365 Copilot, and respond by auditing, warning or blocking. You can simulate a policy before enforcing it.
- ALL
A DAX function that ignores any filters and gives back every row of a table or every value of the named columns. Used within CALCULATE, it works as a modifier that clears filters, although REMOVEFILTERS states that intent more clearly where it is available.
Related terms
- PII
Data that can identify someone, such as their name, email address or national ID number. Unity Catalog can find and label it automatically through Data Classification.