ADA-C01 Question 190
Single answerImplement data classification (EXTRACT_SEMANTIC_CATEGORIES, ASSOCIATE_SEMANTIC_CATEGORIES)A Snowflake administrator needs to classify sensitive data in a newly onboarded CUSTOMER_PROFILE table before applying masking policies. The table contains columns such as EMAIL_ADDR, PHONE_NUM, DOB_TEXT, and NOTES. The administrator wants to use Snowflake's semantic classification workflow so that likely categories are identified from sampled data first, reviewed, and then written back as associations only for approved columns. Which approach best meets this requirement?
- A
Run EXTRACT_SEMANTIC_CATEGORIES on the CUSTOMER_PROFILE table to identify likely semantic categories, review the returned results, and then run ASSOCIATE_SEMANTIC_CATEGORIES only for the approved column-category mappings.
- B
Run ASSOCIATE_SEMANTIC_CATEGORIES on the CUSTOMER_PROFILE table first so Snowflake can infer categories automatically, then use EXTRACT_SEMANTIC_CATEGORIES afterward to validate the applied associations.
- C
Use EXTRACT_SEMANTIC_CATEGORIES to permanently tag columns with classifications, and then use ASSOCIATE_SEMANTIC_CATEGORIES only if you need to remove incorrect classifications.
- D
Create masking policies on all candidate columns first, because EXTRACT_SEMANTIC_CATEGORIES can only analyze columns that already have a masking policy attached.
Show answer and explanation
Correct answer: A
Explanation
Snowflake's semantic classification workflow separates discovery from persistence. EXTRACT_SEMANTIC_CATEGORIES analyzes data and returns likely semantic categories for columns, enabling review and approval. ASSOCIATE_SEMANTIC_CATEGORIES is then used to store the approved semantic category associations on the relevant columns. This staged approach is useful in real environments where administrators need to validate inferred classifications before downstream controls such as masking policies or governance processes rely on them. The key best practice is to avoid automatically persisting all inferred classifications without review, especially for ambiguous columns like free-text NOTES or columns with nonstandard names such as DOB_TEXT. This question reflects Snowflake documentation and recommended governance workflows around semantic category extraction and association.
- A. Correct.
Correct. EXTRACT_SEMANTIC_CATEGORIES is used to analyze data and return recommended semantic categories based on the contents of the specified table or columns. This supports a review step before any metadata is written. ASSOCIATE_SEMANTIC_CATEGORIES is then used to persist approved associations between columns and semantic categories. This is the appropriate workflow when the administrator wants human review before committing classifications.
- B. Incorrect.
Incorrect. ASSOCIATE_SEMANTIC_CATEGORIES does not perform the discovery step by inferring classifications on its own. It is used to associate semantic categories after you have determined which mappings should be applied. Reversing the order defeats the purpose of reviewing extracted recommendations first.
- C. Incorrect.
Incorrect. EXTRACT_SEMANTIC_CATEGORIES does not permanently write classifications to columns. It returns extracted classification results that can be reviewed. ASSOCIATE_SEMANTIC_CATEGORIES is the function used to write associations. It is also not a removal mechanism for incorrect classifications in the way described here.
- D. Incorrect.
Incorrect. Masking policies are not a prerequisite for semantic extraction. In practice, classification is often performed before governance controls such as masking policies are applied, because the classification results help determine which columns need protection. This option reflects a common misunderstanding of the relationship between discovery and enforcement.