PRIVACY PRESERVING RECORD LINKAGE USING PHONETIC AND BLOOM FILTER ENCODING

Now-a-days, there is an increasing demand for data integration and analytics due to the availability of huge amount of records in multiple data sets. In data integration, record linkage gains prominent importance to identify and match records across data sets that belong to the same person. Record linkage becomes complicated with presence of erroneous identifiers and hence needs approximate matching to find similarities between the same person records. Also, the data sharing for record linkage can lead to disclosure of confidential information about the personal records. Thus, Privacy preserving record linkage (PPRL) involves detecting and matching of records among two or more data sets in a secure manner. It is useful for the purpose of research activities and analysis across wide application areas. The utilization of Bloom filter encoding with its hardened versions are suitable for approximate matching in PPRL, but some of them are vulnerable to re-identification attacks while others reduce linkage accuracy. Moreover, phonetic encoding can provide robust matching with its inherent security characteristic for PPRL. However, most of existing PPRL techniques had attempted to provide privacy while compromising the linkage accuracy. This research focuses on designing a new approach named as two factor encoding for PPRL (2FE-PPRL) using phonetic and Bloom filter encoding to achieve increased linkage accuracy while maintaining privacy. Our 2FE-PPRL approach depicts better results than existing PPRL techniques Phonetic and Bloom Filter encoding as analyzed through precision, recall and f-measure.