面向终端用户的组合云服务的提供通常需要多方协作、集成现有资源和服务、并以跨组织的分布式业务流为支撑。然而，云环境是高度动态环境，协作方的服务可能随时发生变化；用户可能随时改变其需求和情景信息。这些变化对支撑应用服务的业务流有不同类型和不同程度的影响。因此，服务提供商需要有完善的变化分析和应对机制，以保证既为用户提供满足服务等级协议的应用服务，又保证协作方的变化对终端用户的透明性，同时，保障自身的收益。然而，现有变化管理方法一般基于以过程为中心的业务流，不适用于当前数据密集的云环境。本项目针对跨组织业务流的变化管理难题，致力于研究两大类变化：协作方关键服务质量参数变化和用户情景信息变化。项目拟构建基于artifact的多层业务流形式化模型，在此基础上，提出变化影响分析的方法和变化应对机制。研究成果将为云环境中面向市场的资源自动分配、系统进化管理、业务流溯源等问题奠定一定的理论基础。

To offer composite cloud services for end users, a service provider needs to collaborate with multiple partners, integrate existing resources and Web services, and manage cross-organizational and distributed business processes. In a highly dynamic cloud environment, variations including functional and nonfunctional changes will happen at any time to a partner's Web services; and users may change their requirements or contextual information as well. All these changes from partners and end users will have different types and levels of impact on the provider's business processes. Thus, to provide application services with satisfied service level agreements, ensure changes to be transparent for end users, as well as guarantee its own profit, a sophisticated approach for change analysis and reaction is required. Existing approaches for change management is based on process-centric workflow models, and they cannot meet the requirements in current data-intensive environments. This project focuses on the challenge problem of change management in the context of cross-organizational collaboration. In particular, we shall study the two major types of changes, i.e. variations on the quality of service at the partner side and the changes of users' contextual information. To manage these changes, we firstly plan to build an artifact-based multilayered business process model. Based on this model, the approach for change impact analysis and reactions will be presented. Results from this research can provide theoretical foundations for the problems of market-oriented automated resource allocation, system evolution management, as well as process provenance.