K-anonymous rule induction from data table of binary classification using imprecise rules

Abstract

While the utilization of open and big data is higly promoted, the privacy protection for personal and sensitive data is indispensable. In this paper, we investigate the anonymization tschniques for rules induced from data tables before the publication. An anonymization method by inducing imprecise rules has been proposed for data tables with not less than three classes. However, for data tables of binary classification, we cannot simnply apply the imprecise rules because of the lack of classes to be combined. Then we invent a way to induce anonymized rules from data tables of binary classification using imprecise rules. We examine the usefulness of the proposed approach.