De-anonymising set-generalised transactions based on semantic relationships

Abstract

Transaction data are important to applications such as marketing analysis and medical studies. However, such data can contain personal information, thus must be sanitised before being used. One popular approach to protecting transaction data is set-based generalisation, where an item in a transaction is replaced by a set of items. In this paper, we study how well transaction data can be protected by this approach. More specifically,we propose de-anonymisation methods that aimto reconstruct original transaction data from its set-generalised version by analysing semantic relationship that exist among the items. Our experiments on both real and synthetic data show that set-based generalisation may not provide adequate protection for transaction data, and about 50% of the items added to the transactions during generalisation can be detected by our method with a precision greater than 80%.