A Hybrid K-anonymity Data Relocation Technique for Privacy Preserved Data Mining in Cloud Computing

The unprecedented power of cloud computing (CC) that enables free sharing of confidential data records for further analysis and mining has prompted various security threats. Thus, supreme cyberspace security and mitigation against adversaries attack during data mining became inevitable. So, privacy preserving data mining is emerged as a precise and efficient solution, where various algorithms are developed to anonymize the data to be mined. Despite the wide use of generalized K-anonymizing approach its protection and truthfulness potency remains limited to tiny output space with unacceptable utility loss. By combining L-diversity and (${\alpha}$,k)-anonymity, we proposed a hybrid K-anonymity data relocation algorithm to surmount such limitation. The data relocation being a tradeoff between trustfulness and utility acted as a control input parameter. The performance of each K-anonymity's iteration is measured for data relocation. Data rows are changed into small groups of indistinguishable tuples to create anonymizations of finer granularity with assured privacy standard. Experimental results demonstrated considerable utility enhancement for relatively small number of group relocations.