Zero-knowledge Test of Vector Equivalence and Granulation of User Data with Privacy

This paper introduces a new framework for privacy preserving computation to the granular computing community. The framework is called P4P (Peers for Privacy) and features a unique architecture and practical protocols for user data validation and vector addition-based computation. It turned out that many non-trivial and non-linear computations can be done using an iterative algorithm with vector-addition aggregation steps. Examples include voting, summation, SVD, regression, and ANOVA etc. P4P allows them to be carried out while preserving users privacy. To demonstrate its application in granular computing, we present two practical protocols that test the equality of user vectors in zero-knowledge. Our protocols only involve constant number of public key operations (independent of vector size) and are very efficient. These protocols can be used to perform granulation, which is a fundamental task of granular computing, in a privacy-preserving manner. They can also be of independent interest for other fields such as data mining as well.