In blockchain networks, cross-chain transaction audit is particularly important due to the wide variety of transactions between different chains, the large number of participants, and the unique rules and mechanisms of each chain. Scoring transactions can help buyers better understand the quality and credibility of transactions. However, the authenticity of the scores is not always reliable. In this paper, we propose a cross-chain transaction auditing scheme to address this problem. To be specific, this scheme deploys smart contracts on the relay-chain to query the seller’s transaction scores on the chain and uses truth discovery technology to evaluate the seller’s reputation. Since the smart contract is open and transparent, the credibility of the calculation process is ensured. In addition, considering the timeliness of the seller’s reputation, we use Dirichlet distribution to predict the future reputation through the seller’s historical reputation truth. Through experimental evaluation, we verified the effectiveness and reliability of the proposed scheme. In a local simulated experimental environment, the experimental results showed that the proposed scheme is feasible and efficient.