<p>Pricing is widely considered as a critical issue for emerging data markets. As a digital commodity, data exhibits several unique characteristics including infinite replicability, zero marginal cost and non-rivalrous nature, which impose significant challenges for traditional data auction models. In data market practice, each auction instance typically has multiple winners, and it is natural and fair for each winner to pay the same price for the same data. However, most of the existing auction mechanisms, such as generalized second price auctions, cannot guarantee this fairness, leading to possible price discrimination in data markets. In this paper, we design a novel uniform second price (USP) mechanism for multi-winner repeated data auctions, which proves to possess approximate incentive compatibility. USP can ensure fairness, market stability, and long-term revenue maximization for sellers. We prove that our USP mechanism can incentivize buyers to truthfully bid their private valuations, and thus can be well adapted to most data trading scenarios. We also investigate the potential existence of a non-truthful bidding strategy space within our USP mechanism through mathematical proofs and empirical simulations, and conduct a comparative analysis validating that the USP mechanism can outperform two prevalent multi-winner auction mechanisms, i.e., second price auction and monopolistic price auction.</p>