SMSPPRL: A Similarity Matching Strategy for Privacy Preserving Record Linkage

Abstract

Now-a-days, huge amount of personal and sensitive data of individuals resides across different data sources that refer to the same entity. Thus, it is crucial and necessary to detect and link duplicate records from multiple data sets in secure manner referred to as privacy preserving record linkage (PPRL). The PPRL enables data integration, analysis and research activities for business benefits. Since real world data exhibits its dirty and erroneous representations, achieving linkage accuracy is a prominent factor for PPRL techniques. Hence, approximate matching techniques play a crucial role for achieving linkage accuracy in PPRL applications. In this paper, different suitable attribute combinations for PPRL are identified. This paper introduces a similarity matching strategy for privacy preserving record linkage named as SMSPPRL for achieving increased linkage accuracy. Our SMSPPRL technique performs better than existing PPRL techniques Basic Bloom, hardened balanced Bloom filter in terms of linkage accuracy.