Woodgrove Bank must detect only the account numbers that belong to its 180,000 real customers, and no other numbers of the same shape. You decide to build an exact data match based sensitive information type. Which two tasks must you complete before the exact data match type can return matches? Each correct answer presents part of the solution. (Choose TWO.)
Choose 2.
- A.
Create a custom trainable classifier that is trained on a sample of the account number table.
- B.
Hash and upload the sensitive information source table by using the EDM Upload Agent, signed in with an account in the EDM_DataUploaders security group.
- C.
Enable optical character recognition for the tenant so that account numbers in images are read.
- D.
Define the exact data match schema, which maps the columns of the sensitive information source table and identifies the primary fields that can start a lookup.
- E.
Publish an auto-labeling policy in simulation mode that references the exact data match type.
Show answer
Answer: B, D
An exact data match type needs a schema that declares which columns are searchable and a hashed, uploaded copy of the sensitive data table before it can match anything.
- A. Trainable classifiers learn a fuzzy content category and are never part of the exact data match build process.
- B. The EDM Upload Agent hashes the table with a salt and uploads only the hashes; the uploading account must belong to the EDM_DataUploaders group, and without an uploaded, indexed table there is nothing to compare content against.
- C. Optical character recognition only extends scanning into images and is not a prerequisite for exact data match detection.
- D. The schema maps the source table's columns and marks which fields are primary (searchable); in the new experience it is generated from a sample file in the same workflow that creates the SIT, but it must exist before anything can match.
- E. Simulation mode is how you test a policy that consumes the classifier; it does nothing to make the exact data match type functional.