Abstract: Data valuation is an essential task in a data marketplace. It aims at fairly compensating data owners for their contribution. There is increasing recognition in the machine learning community that the Shapley value -- a foundational profit-sharing scheme in cooperative game theory -- has major potential to value data, because it uniquely satisfies basic properties for fair credit allocation and has been shown to be able to identify data sources that are useful or harmful to model performance. However, calculating the Shapley value requires accessing original data sources. It still remains an open question how to design a real-world data marketplace that takes advantage of the Shapley value-based data pricing while protecting privacy and allowing fair payments. In this paper, we propose the {\em first} prototype of a data marketplace that values data sources based on the Shapley value in a privacy-preserving manner and at the same time ensures fair payments. Our approach is enabled by a suite of innovations on both algorithm and system design. We firstly propose a Shapley value calculation algorithm that can be efficiently implemented via multiparty computation (MPC) circuits. The key idea is to learn a performance predictor that can directly predict model performance corresponding to an input dataset without performing actual training. We further optimize the MPC circuit design based on the structure of the performance predictor. We further incorporate fair payment into the MPC circuit to guarantee that the data that the buyer pays for is exactly the same as the one that has been valuated. Our experimental results show that the proposed new data valuation algorithm is as effective as the original expensive one. Furthermore, the customized MPC protocol is efficient and scalable.

Pricing Personal Data Based on Information Entropy

A Reasonable Data Pricing Mechanism for Personal Data Transactions with Privacy Concern.

Personal Data Market Optimization Pricing Model Based on Privacy Level

Data Pricing Mechanism Based on Property Rights Compensation Distribution

An Incentive Mechanism for Trading Personal Data in Data Markets

A Theory of Pricing Private Data

How to Balance Privacy and Money through Pricing Mechanism in Personal Data Market

Enhanced User Data Privacy with Pay-by-data Model

A Brief Survey of Data Pricing for Machine Learning

Pricing and disseminating customer data with privacy awareness

Heuristic-Q: A Privacy Data Pricing Method Based on Heuristic Reinforcement Learning

User Privacy Awareness, Incentive and Data Supply Chain Pricing Strategy

Online Pricing with Reserve Price Constraint for Personal Data Markets

Establishing the Price of Privacy in Federated Data Trading

Trading Location Data with Bounded Personalized Privacy Loss

Data Price Determinants Based on a Hedonic Pricing Model

A Survey on Data Pricing: from Economics to Data Science

Locally Differentially Private Personal Data Markets Using Contextual Dynamic Pricing Mechanism

Private Data Valuation and Fair Payment in Data Marketplaces

Pricing Personal Data Based on Data Provenance

A Survey of Data Pricing for Data Marketplaces