A Partial Deletion Strategy of Set-valued Data Anonymization
XU Xin-hu · Jisuanji gongcheng · 2013
Privacy-preserving under set-valued data publishing is an important problem. Aiming at this problem, this paper presents an iterative strategy that anonymizes set-valued data through partial deletion strategy. This strategy ensures that no strong inferences of sensitive information are possible regardless of the amount of background knowledge the attacker possesses, while making no particular assumption of the downstream utility of the data. It attempts to retain as many mineable useful association rules as possible in the anonymized data, while minimizing the item deletions. Experimental result shows that partial deletion significantly outperforms generalization and global deletion, two of the existing popular anonymization techniques, reducing the number of deletions by 30% on average and retaining 25% more rules.