A Fast Algorithm for Privacy-Preserving Utility Mining

Abstract

Utility mining (UM) is an efficient technique for data mining which aim to discover critical patternsfrom various types of database. However, mining data can reveal sensitive information of individuals. Privacy preserving utility mining (PPUM) emerges as an important research topic in recent years. In the past, integer programming approach was developed to hide sensitive knowledge in a database. This approach required a significant amount of time for preprocessing and formulating a constraint satisfaction problem (CSP). To address this problem, we proposed a new algorithm based on a hash data structure which performs more quickly in itemsets filtering and problem modeling. Experiment evaluations are conducted on real world and synthetic datasets.