社会依赖演化网调控的Agent服务协同自适应

The service-oriented reformation of application software systems has already facilitated service cooperation (i.e., service-based cooperation) to become the mainstream mode for innovating distributed cooperation software systems. However, the new-type software systems based on service cooperation are faced with the great difficulty for configuring self-adaptation capability, and this makes those systems not to be deployed on a large scale in complex and dynamically changing Internet computing environments. In order to enable the service cooperation and software systems based on service cooperation to acquire the robust and high-efficient capability for self-adaptation, this project proposes to develop the agent service cooperation self-adaptation governed by social dependence evolution networks. The research objects include creating the architecture for governing cooperation processes based on social dependence evolution networks, and innovating traditional centrally-controlled and closed self-reflection computing into the open united one governed by social dependence evolution networks. The former closely couples the three aspects of technologies in inter-support: the cooperation process regulations represented as social dependence evolution networks, rational agents, and the cooperation supervising system in community, to remove the hindrances which the cooperation self-adaptation is confronted with, and to create the solid foundation for achieving cooperation self-adaptation, while the latter establishes the cooperation circumstance-driven mechanism for monitoring service cooperation process and the management policy-driven 4-phase circulation for self-adaptation, in order to enable the cooperation processes and their self-adaptation to be executed compactly and consistently under the government of social dependence evolution networks.

应用软件系统的服务化变革促使服务协同(基于服务的协同)成为革新分布式协同软件系统的主流方式，但难以配置自适应能力又导致基于服务协同的新型软件系统无法大规模部署于复杂、动态变化的Internet计算环境。为使服务协同和基于服务协同的软件系统能够获得鲁棒的高效用自适应能力，本项目提议开发社会依赖演化网调控的Agent服务协同自适应，包括建立基于社会依赖演化网的协同过程调控架构和革新传统的集中控制型封闭性反思计算成为社会依赖演化网调控的开放性联合反思计算。前者紧密耦合相辅相成的三方面技术：表示为社会依赖演化网的协同过程法规、理性Agents和社区协同监管体系，去消除协同自适应面临的障碍：跨管理域业务服务固有的不可控制性，并给协同自适应的实现奠定坚实基础；后者则建立协同情境驱动的协同过程联合监控机制，以及管理政策驱动的自适应循环，使得协同过程及其自适应能够在社会依赖演化网的调控下紧凑、一致地执行

应用软件系统的服务化变革促使服务协同(基于服务的协同)成为革新分布式协同软件系统的主流方式，但难以配置自适应能力又导致基于服务协同的新型软件系统无法大规模部署于复杂、动态变化的Internet计算环境。为使服务协同和基于服务协同的软件系统能够获得鲁棒的高效用自适应能力，本项目开发了社会依赖演化网调控的Agent服务协同自适应，包括建立基于社会依赖演化网的协同过程调控架构和革新传统的集中控制型封闭性反思计算成为社会依赖演化网调控的开放性联合反思计算。..通过建立基于社会依赖演化网的协同过程调控架构，使得服务协同过程及其包含的协同活动转变为可预言和可控的，从而是可信的。由此，不仅可以克服跨管理域协同活动的不可控障碍，也为服务协同应用的部署消除了信任危机。同时，使得企业代理Agents在按需动态建立服务协同关系和企业联盟后，能够随即投入自主执行，并自动形成对于协同活动的联合监控, 实现了服务协同应用期望但传统软件工程的集中控制模式无法达到的、服务协同应用投入运行的及时性。.通过建立基于调控架构的三层次Agent服务协同自适应机制，实现了社会依赖演化网（协同过程模式）调控的开放性联合反思计算，去革新传统的集中控制型封闭性反思计算；使得协同开展过程及其自适应能够遵循演化网指定的协同状态变迁和依赖关系演化（有条件激活的规范相继执行）来实现。同时，通过这三个层次自适应活动的相辅相成，不仅有效降低了实现自适应的复杂度和难度，也进一步提高了实现自适应的有效性和高效性。