L diversity

The function used to check for l-diversity in a table

Identity disclosure - Link a record with the person's identity, eliminated using k-anonymity

Attribute disclosure - Gather more information about the person with respect to the sensitive attributes, can be eliminated by l-diversity and t-closeness, is an implication of k-anonymity

Attacks on a k-anonymized dataset

• Homogenity attack: given an equivalence class, if the values of the sensitive attributes are not diverse enough, it is easy for the attcker to conclude the value of the sensitive attribute of the subject
• Background knowledge attack, if the attacker knows something about the subject's sensitive attribute, it could be derived from looking at the tavble

we showed that k-anonymity is susceptible to homogeneity and background knowledge attacks

Background attack

Know record is in the table

Know the qids and their domains

Know something about sensitive attribute and how it is related to the person