A Survey on Classification over Semantically Secure Encrypted Data

Abstract — with the rapid development of web services and their popularity, web users are increasing day by day. As a result, there is a large and heterogeneous data. This data needs to be mine for various real time and other type of applications like banking, medicine, scientific research and among government agencies. Data Mining is a wider area now days due to the necessity of knowledge discovery from a different perspective on a very large scale. One of the commonly used tasks in data mining applications is the Classification. Many theoretical and practical solutions to the classification problem have been proposed in the past decades. To overcome the privacy issues, these solutions have used different security models. The current needs of IT makes the Cloud Computing comes into existence. Users can outsource their data in encrypted form and the data mining tasks to the cloud. Existing privacy preserving classification techniques are not applicable to the encrypted data on cloud. Therefore, this paper aims to solve the classification problem on encrypted data. A k-NN classifier over encrypted data in the cloud is proposed here for security purpose. The system proposes a protocol to provide the confidentiality of the data on the cloud protects privacy of user input query and hides the data access patterns on the cloud. A reliable k-NN classifier is the first over the encrypted data under the semi-honest model