DPPDI: Efficient Distributed Privacy-Preserving Data Integration for Large Datasets
Jiaer Jiang, Jinjiang Yang, Jingcheng Zhao, Yingjie Xue, Kaiping Xue · 2025
Privacy-preserving data integration (PPDI) is a secure method to integrate datasets from different data sources while protecting the privacy of data. Existing PPDI work usually uses the outsourced framework and executes data integration through a cloud server. Due to the need to protect the privacy of the relations between IDs and associated data, the associated data must be encrypted or blinded before uploading to the cloud server, which leads to poor performance. For the efficient PPDI solution, we first carefully analyze the privacy goals of PPDI. After that, we adopt the distributed computing model, and then propose a multi-party PPDI protocol named DPPDI. Our scheme removes the overhead caused by encrypting associated data while protecting privacy, and realizes the outer join functionality and arbitrary combination of data sources. Besides, to avoid dropping records when duplicate IDs exist, we propose a method embedded into the PPDI protocol to handle duplicate IDs. Finally, we conduct extensive experiments to evaluate our scheme's performance, and the result shows that our scheme outperforms previous PPDI schemes.