Privacy-preserving credit evaluation system based on blockchain

Abstract

In digital intelligence era, the authenticity of data and the privacy protection of data sharing and multiparty collaborative computing are key factors in building a good credit evaluation system. Although blockchain-based credit evaluation system considered this point, more studies on data privacy protection are either only the design of the framework or the proposal of the concept, which is insufficient for privacy protection. To provide a more comprehensive and reliable privacy-preserving scheme, this paper proposes a novel credit evaluation system with secure sharing and multiparty computation based on blockchain. The system consists of five modules: data, access control, data encryption, secure computation and model storage modules. The Hyperledger Fabric blockchain-based data module ensures the authenticity and traceability of the data source. The raw data are encrypted by a linear transformation algorithm in the data encryption module, which minimizes the output and utilization of data when leaving local storage and prevents potential privacy leakage in data sharing to the greatest extent. The phillie homomorphic encryption-based secure computation algorithm in the secure computation module achieves secure data sharing while applying the secure multiparty computation, which makes it possible to data sharing and privacy protection in multiparty computing. The system obtains final summary statistical results without exposing the raw data. Additionally, the final statistical results of the raw data can be inferred from the encrypted data, and their results are consistent. The correctness and accuracy of the linear conversion encryption mechanism and homomorphic encryption algorithm are proven by theoretical analysis. Security analysis and calculation case show the security of the proposed credit evaluation system and the correctness and effectiveness of the proposed encryption scheme.