
464 Analyzing and Securing Social Networks
34.3.3 l-Diversity
l-Diversity (Machanavajjhala et al., 2006) ensures that there are at least l well-represented val-
ues of the attributes for every set of quasi-identier attributes. The weakness is that one can
still estimate the probability of a particular sensitive value. m-Invariance (Xiao and Tao, 2007)
ensures that each set of quasi-identier attributes has at least m tuples, each with a unique set
of sensitive values. There is at most 1/m condence in determining the sensitive values. Others
enhanced the techniques of k-anonymity and l-diversity with personalization, such as personal-
ized anon ...