Transparent Link: A Framework of Anonymizing MSA-Dataset Based on Probabilistic Graphical Model
Li-e Wang, Yameng Guo, Xianxian Li · 2017
Privacy preserving of multi-sensitive attributes datasets (MSA-Datasets) has received increasing attention because of its huge social and economic benefits. In this paper, we introduce a novel and general privacy framework called Transparent Link. The Transparent Link framework can be used to anonymize MSA-Datasets by designing an algorithm based on probabilistic graphical model, which is referred to as APGM. Under the framework, to privately protect the relationships among multiple sensitive attributes, we present a clustering approach which can improve the utility of association rules through probabilistic edge association based on multipartite graphs. Experimental results show that our approach offer strong tradeoffs between privacy and utility.