Question 29 — AB-650 Practice Exam 1
A Microsoft Purview administrator must classify a database export containing a known set of customer identifiers. Many unrelated documents contain strings that resemble the identifier format, and the administrator wants to minimize false positives by flagging only identifiers that belong to the approved dataset. What should the administrator use? (Select one answer)
A trainable classifier trained with sample database exports
A sensitive information type that detects the identifier format
Exact data match using the known identifier dataset
A custom sensitive information type using supporting keywords
Correct answers
Exact data match using the known identifier dataset
Explanations
Correct:
Exact data match using the known identifier dataset
EDM verifies candidate values against a known reference dataset, so a value is classified only when it corresponds to an entry in that dataset. This directly addresses the requirement to distinguish genuine identifiers from unrelated strings with a similar appearance.
Wrong:
A trainable classifier trained with sample database exports
Trainable classifiers recognize broader contextual characteristics learned from examples. They are suited to complex or free-form content rather than validating individual values against a fixed enumerable list.
A sensitive information type that detects the identifier format
A pattern-based SIT can determine that a string has the expected structure, but it does not establish that the value belongs to the organization's approved dataset. Plausible-looking unrelated values can therefore produce false positives.
A custom sensitive information type using supporting keywords
Supporting keywords can strengthen pattern detection, but they still do not provide exact membership validation against the known set of identifiers. EDM is designed for that requirement.