At present most privacy preserving algorithms based on l-diversity model are limited only to static data release. It is low efficiency and vulnerable to inference attack if these anonymous algorithms are directly applied to dynamic data publishing. To address this issue, this paper analyzes various inference channels that possibly exist between multiple anonymized datasets and discusses how to avoid such inferences and provides an effective approach to securely anonymize a dynamic dataset based on incremental clustering: incremental l-diversity algorithm. Theory analysis and experiment results show that the proposed method is effective and efficient.
Wang, Pingshui and Wang, Jiandong
"L-Diversity Algorithm for Incremental Data Release,"
Applied Mathematics & Information Sciences: Vol. 07
, Article 46.
Available at: https://dc.naturalspublishing.com/amis/vol07/iss5/46