A database is reported as "3-anonymous, 3-diverse, 0.167-close for salary, 0.278-close for disease." What does each number mean?

Each metric reports a different privacy guarantee: group size (k=3), value variety (l=3), and how closely each sensitive attribute's distribution matches the whole (t via EMD).

Reading the report card:

• 3-Anonymous (k=3): every equivalence class has at least 3 indistinguishable records (group size / identity protection).
• 3-Diverse (l=3): every class contains at least 3 distinct sensitive values (variety / attribute protection).
• 0.167 close for Salary (Numerical EMD): the worst class's salary distribution is within an Earth Mover's Distance of 0.167 of the global salary distribution.
• 0.278 close for Disease (Categorical EMD): likewise for the disease attribute, using a categorical EMD.

Together these mean no specific sensitive information can be reliably inferred — the three models stack to close identity and attribute leaks.

Tip: k, l, and t are three different rulers. A serious anonymization report states all three, because each closes a different attack.