In the analysis of information loss, it is found that with the number of identifiers increasing in the Magic and EIA datasets, the algorithm in this paper is significantly lower than that in k-anonymity and l-diversity models, and achieves the expected effect. In the availability scoring analysis, for both Magic data set and EIA data set, the usability improvement value of k-anonymity-differential privacy algorithm proposed in this paper is the most significant when the ? value is 0.1. In the operation efficiency analysis, the algorithm constructed in this paper is the most efficient in the Magic data set.
Zhihan Lv, F. Piccialli
ACM Transactions on Internet Technology (TOIT)