Credit system is generally associated with the banking and financial institutions, although it has far reaching implications for residents of countries, such as U.S., particularly for those with a poor credit history. Specifically, a credit score computation (CSC) quantifies an individual’s credit value or credit risk, which is used by banking and financial institutions, as well as other entities (e.g., during purchasing of insurance policies and application of rental properties), to facilitate their decision-making (e.g., whether to approve the insurance policy purchase or the level of premium). Although a number of CSC models have been proposed in the literature for supporting different application scenarios, privacy protection of CSC is rarely considered despite the potential for leakage of user private information (e.g., registration, hobbies, credit, relationships, and inquiry). Such information can then be abused for other nefarious activities, such as identity theft and credit card fraud. Thus, in this article, we first analyze the privacy strength of existing CSC models, prior to presenting the formal definition of a privacy-preserving CSC system alongside its security requirements. Then, we propose a concrete construction based on Paillier encryption with three proposed noninteractive zero-knowledge schemes. To demonstrate feasibility of our proposal, we evaluate both its security and performance.
Read full abstract