Comparison on membership functions in fuzzy k-member clustering for data anonymization

k-member clustering is an efficient method of k-anonymization, in which data samples are anonymized so that any sample is indistinguishable from at least k-1 other samples. Fuzzy k-member clustering is a fuzzy variant of k-member clustering, which extracts k-member clusters with fuzzy memberships of samples and makes it possible for the samples having large residual memberships to belong to second or later clusters. By allowing boundary samples to be shared by multiple clusters, data anonymization is performed without significant loss of information. In this paper, several shapes of membership functions used in the calculation of the fuzzy memberships are compared from the view point of information loss in anonymization.