Efficiently Verifiable Computations on User Data for Behavior Analysis with Privacy Preservation

Abstract

Along with development of internet and web, online social network are becoming important information propagation platform with hundreds of million users worldwide. Online social network attract thousands of million users to use it every day for different purpose. So that tons of user behavior data is generated on internet. Developing endeavors have been committed to mining the inexhaustible behavior data to extract significant information for research purposes to  inquire about that, or analyst to develop better ecommerce strategies for business purpose. However the concern arises with this data is security, which is going to be presented to third parties. The most recent decade has seen an assortment of look into works endeavoring to perform information conglomeration in a privacy protecting manner. Most by far of existing  techniques give protection to users information yet at the cost of very limited data aggregation operations like calculating sum and mean of particular query, which barely fulfill the requirement of behavior analysis. So that, proposed system mainly focuses on privacy preservation and behavior analysis of online user data. In this paper we use general accumulation and specific collection for behavior analysis. Using cryptographic algorithm we prevent privacy disclosure from both  third party data aggregator and analyst.  We have executed our technique and assessed its execution utilizing a relational dataset. The results of the experiment shows that this research scheme handle both overall queries and various selective aggregate queries  with acceptable computation, privacy, and  overheads of the communication effectively.