Proposal for New PSI Protocol in PPDP Environment

In secure data sharing and collaborative data analysis, the Private Set Intersection (PSI) Protocol is a key protocol that enables multiple data users to securely find common elements in their datasets without revealing the datasets themselves.Ongoing research in this area focuses on various implementation methods.Pinkas proposed a Bloom filter-based Private Set Intersection Protocol 1, which is effective in terms of computational complexity but faces challenges in effectively reducing false positives.To address this issue, the Garbled Bloom Filter approach was suggested; however, it exhibited problems such as increased computational complexity and memory usage.Improved hash functions were proposed in Cuckoo Hashing to mitigate overflow caused by collisions during data input.Among these, the Minimum Perfect Hashing, which adheres to the one-to-one correspondence principle between hash table indices, was suggested.However, it only applies to static datasets and incurs additional data addition or deletion costs.In this paper's Privacy-Preserving Data Publishing (PPDP) environment, a PSI protocol was proposed using Oblivious Transfer-based minimal perfect hashing that achieves zero false positives with high memory efficiency and fast query performance.Furthermore, its parallel processing capability allows for its application to dynamic datasets without additional computational costs.