$\ell\rho$-suppression: A Privacy Preserving Data Anonymization Method for Customer Transaction Data Publishing
Dedi Gunawan, Yusuf Sulistyo Nugroho, Fatah Yasin Al Irsyadi, Ihsan Cahyo Utomo, Ilham Andreansyah, Syful Islam · 2022
Publishing a transaction database to other parties is becoming a common way to harvest useful information from the database. Though, the database usually contains privacy concerns such as personal private itemset. Therefore, it is crucial to protect sensitive information prior to sharing the database to other parties. A solution to preserve the privacy of data subject in the transaction database is by modifying the transaction records using specific data anonymization method. Data anonymization method such as local generalization is widely adopted to solve the issue. However, the method induces excessive item loss and reduce data utility significantly. In this paper, a privacy preserving method to protect private itemsets of individual data subjects namely$\ell\rho$-suppression is proposed. The method adopts cell suppression technique where personal sensitive items in transaction records are omitted. The method is distinct from an existing method, where it uses maximum term frequency normalization of the items in the database and it only suppresses items from transaction records having the privacy ration higher than that of$\ell\rho$value. Experimental results indicate the proposed methods can successfully enhance the privacy protection level and has the ability to minimize the number of item loss as well as maintaining data utility level in an anonymized database.