Accurate and efficient privacy-preserving string matching

Abstract

The task of calculating similarities between strings held by different organisations without revealing these strings is an increasingly important problem in areas such as health informatics, national censuses, genomics, and fraud detection. Most existing privacy-preserving string matching approaches are either based on comparing sets of encoded characters allowing only exact matching of encoded strings, or they are aimed at long genomics sequences that have a small alphabet. The set-based privacy-preserving similarity functions that are commonly used to compare name and address strings in the context of privacy-preserving record linkage do not take the positions of sub-strings into account. As a result, two very different strings can potentially be considered as a match leading to wrongly linked records. Furthermore, existing set-based techniques cannot identify the length of the longest common sub-string across two strings. In this paper, we propose two new approaches for accurate and efficient privacy-preserving string matching that provide privacy against various attacks. In the first approach we apply hashing-based encoding on sub-strings (q-grams) to compare sensitive strings, while in the second approach we generate one-bit array from the sub-strings of a string to identify the longest common bit sequences. We evaluate our approaches on several data sets with different types of strings, and validate their privacy, accuracy, and complexity compared to three baseline techniques, showing that they outperform all baselines.

Fraud detection through data sharing using privacy‐preserving record linkage, digital signature (EdDSA), and the MinHash technique: Detect fraud using privacy preserving record links

“…Vaiwsri et al. [6] introduced an implementation without Bloom filters in which hashing‐based encoding is applied to the q ‐grams and longest common bit sequences to identify a match. This implementation has a semi‐trusted third‐party linkage unit to perform matches and inform the respective parties.…”

“…Their cryptanalysis even overcomes some of the Bloom filter hardening techniques. Usage of third parties: Implementations such as the one by Vaiwsri et al. [6] use a semi‐trusted third party to perform matches. Third parties cannot be used in the Telecom industry due to privacy laws.…”

“…Furthermore, telcos may not agree on a mutually trusted third party. Using a third party also involves multiple transactions between parties to determine a match, as shown by Vaiwsri et al [6]. • Complex transactions and speed: Many implementations process data into q-grams, extract the frequency of characters and hamming distance, and pseudo-randomize the data to have additional layers of security.…”

“…Also, many PPRL implementations use Bloom filters [4]. These implementations are prone to frequency attacks and are not secure, as per Schnell [5] using various cryptanalysis techniques over the q ‐grams of a Bloom filter. Many use a trusted third party to transact secured attributes [6]. This technique is employed to ensure that the two parties do not reveal their information. They are complex to implement, slow, and have multiple transactions.…”

Privacy-preserving record linkage using autoencoders

“…The lack of common unique entity identifiers (such as social security numbers or patient identifiers) across the databases to be linked means that linking records is commonly based on available quasi-identifiers (QIDs), such as the names, addresses, and dates of birth of the individuals whose records are to be linked [7]. Given these are personally identifiable information [26], concerns about privacy and confidentiality limit or even prevent such personal data from being used for the linkage of records across databases [16,40].…”