Federated $K$-Dominant Skyline: an Efficient Approach under Differential Privacy

Guoqing Cai,Yexuan Shi,Hao Zhou,Nan Zhou,Manxue Guo,Zimeng Jia

DOI: https://doi.org/10.1109/docs60977.2023.10294954

2023-01-01

Abstract:$k$ -dominant skyline queries have been widely used in many applications, such as multi-criteria decision making and recommendation. In recent years, a new computing paradigm, called data federation, has emerged to provide collaborative data analytics among several data owners who would like to share their data under privacy protections. This paradigm brings a new challenge to the $k$ -dominant skyline query, where data is distributed among these owners and privacy leakages should be prevented during the query processing. Accordingly, we define the problem of federated $k$ -dominant skyline query and show that existing solutions cannot be extended to answer this query with satisfactory efficiency. To improve the efficiency, we present an observation of a vector aggregation based re-formulation and design an efficient solution under differential privacy. Finally, extensive experiments are conducted to demonstrate that our solution is notably faster than the state-of-the-arts.