中文摘要：

提出了一个基于移动Agent的服务组合框架（MAWSCF）。引入计划（P），用于描述Agent的逻辑行为，同时给出了过程匹配规则，将组合规范转换为P。在运行过程中，行为匹配算法将P解释为Agent的物理行为，从而实现组合服务任务。MAWSCF使用对等计算（P2P）的模式克服了传统服务组合系统中心控制的性能瓶颈，Agent通过克隆能够并行的执行目标组合任务中的多个分支，大大提高了效率，也有效地降低了网络的负荷。

英文摘要：

A mobile-agent-based web service composition framework （MAWSCF） is proposed. In this paper, ＂Planning （P）＂ is defined to specify the logic actions of mobile agent and the process mapping rules are given to convert the specification to P. During the running time, P is explained as the behaviors of agents by matching algorithms, so that the task described by the specification is achieved correctly. Compared with the traditional WSC Models, MAWSCF avoids the performance bottleneck of center peer by peer-to-peer （P2P） approach. Through cloning, mobile agents achieve the sub-tasks in parallel, which raises efficiency greatly.