Cross-chain swaps are a crucial application that facilitates the transfer of digital assets across different blockchains, thereby enhancing the flexibility and availability of asset circulation. Ensuring atomicity is a fundamental objective for cross-chain swaps. The Hashed TimeLock Contract (HTLC) protocol is one of the primary solutions for cross-chain swaps. It ensures atomicity by statically partitioning execution actions for swap submission and asset refund along the time dimension. Such designs rely on a predictable upper bound on transaction confirmation time. However, this assumption does not hold in practical blockchain environments, leading to atomicity violations. To address this limitation, we propose Chuchu, a hashlock group protocol for cross-chain swaps. Chuchu abandons static partitioning along the time dimension and instead distinguishes swap submission and asset refund through explicitly defined asset locking states and their dynamic transitions. To cope with divergent execution progress across blockchains, Chuchu bounds execution progress divergence and introduces an exit-proof mechanism, ensuring that locked assets always have well-defined and consistent unlocking paths under all execution scenarios. The effectiveness and feasibility of Chuchu are demonstrated through theoretical analysis and formal verification using TLA+. Meanwhile, experiment results show that Chuchu reduces execution time by over 98% compared to the HTLC protocol in the case of swap rollbacks.

Hashed Timelock Contract (HTLC) is a widely-used protocol for cross-chain asset swaps. However, it relies on serial asset-locking to guarantee atomicity, which causes high latency and poor fairness. Aiming at the drawbacks of HTLC, we propose Lilac, a cross-chain asset swap protocol that supports parallel asset-locking. Lilac replaces the unique asset-unlocking credential in HTLC with multiple sub-credentials generated by all participating users, and the sequence of sub-credentials is used as the complete asset-unlocking credential. Users obtain the complete credential only when all assets have been locked, and the credential construction process is independent of the order in which assets are locked, so atomicity can be guaranteed when users lock their assets in parallel. Experiments show when a swap involves 2 to 4 blockchains, Lilac reduces the swap latency by 36.75% to 62.20%. Moreover, Lilac reduces the waiting time gap between different users so the fairness of a swap is improved.

The research results from the attribution theory, equity theory and expectation theory in service recovery have heavy impacts on understanding customer satisfaction with service recovery performance. In this work, the causal relations among these theories are analyzed with influence diagram. Finally a new customer satisfaction mode of service recovery is proposed with the combination of structural equation modeling method. Empirical evidences show that the new model provides a rational method in analyzing customer satisfaction with service recovery, and it is also a helpful tool for service provider in controlling the service recovery quality.