🤖 AI Summary
This paper addresses privacy-preserving record linkage (PPRL): the problem of accurately identifying record pairs across distributed databases that refer to the same entity, while provably protecting sensitive raw data. To this end, we propose a novel PPRL method based on *q*-gram embedding and binary encoding. Specifically, *q*-grams are first mapped into a low-dimensional semantic embedding space; subsequently, a learnable quantization strategy transforms these embeddings into compact binary codes, enabling efficient and cryptographically secure similarity computation. The approach significantly improves matching accuracy—particularly for short-string records—and enhances robustness against re-identification and inference attacks. Extensive experiments on multiple real-world datasets demonstrate an average 8.2% improvement in F1-score over state-of-the-art baselines, while satisfying rigorous formal privacy guarantees (e.g., differential privacy or cryptographic security, as instantiated).
📝 Abstract
To discover new insights from data, there is a growing need to share information that is often held by different organisations. One key task in data integration is the calculation of similarities between records in different databases to identify pairs or sets of records that correspond to the same real-world entities. Due to privacy and confidentiality concerns, however, the owners of sensitive databases are often not allowed or willing to exchange or share their data with other organisations to allow such similarity calculations. Privacy-preserving record linkage (PPRL) is the process of matching records that refer to the same entity across sensitive databases held by different organisations while ensuring no information about the entities is revealed to the participating parties. In this paper, we study how embedding based encoding techniques can be applied in the PPRL context to ensure the privacy of the entities that are being linked. We first convert individual q-grams into the embedded space and then convert the embedding of a set of q-grams of a given record into a binary representation. The final binary representations can be used to link records into matches and non-matches. We empirically evaluate our proposed encoding technique against different real-world datasets. The results suggest that our proposed encoding approach can provide better linkage accuracy and protect the privacy of individuals against attack compared to state-of-the-art techniques for short record values.