IMPROVISATION OF K-NN CLASSIFIER ON SEMANTICALLY SECURE ENCRYPTED RELATIONAL DATA

By the rapid improvement in web help and their popularity, web customers are developing day by day. Hence, there is large and various data. Data Mining has a wide use for the fields of business, medicine, experimental research and among government offices. One of the generally used tasks in data mining applications is Classification. Various professional and possible solutions to the classification problem have occurred introduced in the earlier decades. To overcome the privacy problems, certain clarifications have used various security types. Customers can outsource their data onto encrypted information and the data mining tasks to the cloud. Current privacy preserving classification techniques are not suitable for this encrypted data over the cloud. Securing proper privacy and protection of the data stored, transmitted, prepared, and distributed among the cloud as well as from the users entering such data is one of the big challenges to our current society. Hence, this paper proposes to define the classification problem of encrypted data. A k-NN classifier over encrypted data onto the cloud is proposed here for security reason. This method proposes a protocol to implement the confidentiality of the data onto the cloud protects the privacy of user input query and hides the data access patterns of the cloud. A certain k-NN classifier is the first above the encrypted data onto the semi-honest form. Since developing the performance of SMINn is an essential first step for improving the performance of our PP-k-NN protocol, the alternative and more effective solution than SMINn is studied that extends to different classification algorithms