Novel Approach of Data Transformation for Privacy Assurance using Optimization of Genetic Algorithm

Abstract

Data mining techniques were used to extract important information from the data. There is an inherent risk to the data's privacy when taking data mining. Privacy preservation in data mining process employs approaches that operate on sensitive data that is hidden even from the algorithm's operator. The majority of privacy-preserving approaches rely on reducing the granularity of data representation. This results in data loss, but it improves protection. As a result, there is a trade-off in PPDM between information loss and privacy. Effective techniques that do not jeopardize security mechanisms are needed. Randomization approach, k-anonymity model, l-diversity, and distributed privacy preservation are some of the strategies suggested for privacy preservation. The k-anonymity model is based on a quasi-identifier, which is a set of database attributes that serves as the data's identifier. All of the data in the database is supposed to be organized into a series of tables, with each tuple containing information about one of the tables. K-anonymity strategies depend on the use of pseudo-identifiers to reduce granularity in data representation.. The actual value of the attribute is entirely eliminated using the suppression process. However, these two approaches result in a loss of some information, which can compromise accuracy. In this paper, we proposed a optimization of genetic algorithm for privacy preservation in data mining process.