Privacy Preserving Optimized Fuzzy like Search over Encrypted Data using Phonology

With increasing need of e-Governance for boosting principle of "Minimum Government -Maximum Governance", lots of government software applications were developed to capture citizen"s data and deliver various services to them. Sometimes captured data (like KYC data) is highly sensitive in nature and governed by personal information protection laws. Data at rest is of no use unless it is analyzed to generate useful information. Since the ownership of data lies with different government bodies, analysis over this data was possible only in isolation or else data privacy is compromised on data sharing. One approach to maintain data privacy before sharing it is to encrypt it using an encryption technique but strong Encryption techniques use diffusion principle identified by Claude Shannon which makes analysis over Cipher text inefficacious. This is major deterrent in discovery of useful patterns, relationships and insights in contrast to when analysis is done over multiple data sources collectively which can be important input for Government Agencies in thwarting untoward incidents. This paper presents an approach to make use of Double Metaphone algorithm which reduces English words to their phonetic representations by using combination of consonant sounds and plausible pronunciations thereby achieving efficient fuzzy like search using exact match over encrypted phonetic representations. The string variations created by typo mistakes during search by a user was handled by the phonology and consonant substitution, thereby achieving speedy and accurate fuzzy like search over encrypted data. This not only preserves the privacy but enables data owners to share the data to the cloud for running further analytics over cipher text and sharing encrypted results with authorized government agencies on demand for further decryption with pre-shared symmetric key.