摘要： 跨链桥作为一种新型区块链基本服务，实现了区块链间的资产和数据流通，极大地促进了区块链去中心化金融的建设。然而，在实际应用过程中，由于跨链桥的交易不透明与资产难追踪问题，许多犯罪分子开始将跨链技术作为一种洗钱手段，在链间进行非法资产的转移。因此，跨链交易追踪问题正逐渐成为学术界与相关监管部门关注的焦点。现有方法主要通过规则匹配的启发式算法来实现跨链交易的关联，这种方法虽能有效地匹配特定区块链间的交易，但仅面向中心化金融（Centralized Finance，CeFi）跨链桥服务，且高度依赖于中心化的内部应用程序编程接口（Application Programming Interface，API），难以广泛支持多种跨链桥的交易关联。提出一种面向去中心化金融（Decentralized Finance，DeFi）跨链桥服务的交易关联方法解决以上问题。根据跨链桥消息传递机制获取源链的存款事件日志，并基于专家规则提取关键跨链信息。利用数据爬取技术构造目标链交易搜索空间。在此基础上，通过多种业务规则实现跨链交易的关联。该方法在Celer cBridge、Multichain和Poly Network三座跨链桥的真实交易数据集（以太坊作为源链，币安智能链作为目标链）上进行了验证，准确率分别达到了93.6%、95.3%和91.6%，具有较好的关联效果。

Abstract: As a new type of basic blockchain service, cross-chain bridge realizes the circulation of assets and data between blockchains, and greatly promotes the construction of blockchain decentralized finance. However, in the process of practical application, due to the opacity of cross-chain bridge transactions and the difficulty of tracking assets, many criminals begin to use cross-chain technology as a means of money laundering and transferring illegal assets between chains. Therefore, the issue of cross-chain transaction tracking is gradually becoming the focus of academic circles and relevant regulatory authorities. Existing methods mainly achieve cross-chain transaction association through heuristic algorithms of rule matching. Although this method can effectively match transactions between specific blockchains, it is only oriented to Centralized Finance (CeFi) cross-chain bridge services, and is highly dependent on the centralized internal Application Programming Interface (API), making it difficult to broadly support multiple cross-chain bridges for transaction association. A transaction association method for Decentralized Finance (DeFi) cross-chain bridge service is proposed to solve the above problems. Obtain the deposit event log of the source chain based on cross-chain bridge messaging mechanism and extract key cross-chain information based on expert rules. Construct the target chain transaction search space using data crawling technology. On this basis, the association of cross-chain transactions is realized by multiple business rules. The method is validated on real transaction datasets (Ethereum as the source chain and Coin Smart Chain as the target chain) from three cross-chain bridges, Celer cBridge, Multichain and Poly Network. The accuracy rates reach 93.6%, 95.3%, and 91.6%, respectively, with high association accuracy.