服务互联网（Internet of Service ）环境下，服务消费者在高效率地利用海量业务服务构建服务组合的过程中，仍然面临信息过载、动态演化的服务组合模式、大规模个性化的客户需求等困难。传统的服务组合方法在用户需求建模和服务组合生成方面仍存在缺乏推理能力、依赖人工介入的问题，难以适应快速响应用户需求、自动生成服务组合的需要。本项目拟选择旅游服务组合，即旅行包为具体背景，面向自然语言形式的用户请求，考虑服务组合模式的动态演化，研究如何自动构建并推荐服务组合。研究内容包括四方面：首先，研究自然语言形式用户请求的需求分解方法，并自动生成服务组合方案；然后，考虑服务组合模式的动态特性，研究服务组合方案集合的演化机理与生成方法；第三，对用户请求及上下文信息进行获取和建模，研究融合上下文与用户请求的服务组合推荐方法；最后，通过构建基于SOA架构的智能旅游服务平台原型系统来对研究的成果进行验证。

Under the environment of Internet of Service, the service consumers still need to face the issues of severe information overload, dynamically evolutionary pattern of service composition, and large-scale personalized queries, to reuse the massive business services effectively and build service compositions for value-add. Traditional methods for service composition lack the ability of reasoning and still need a lot of manual intervention when building profiles for the requirements of consumers and making service compositions, thus cannot fit the demand of quick response to user queries and generation of service compositions. In this project, the travel service composition, i.e., travel package, is chosen as a specific background for studying, and we try to study how to build service compositions with high quality automatically for the large-scale personalized queries, with the dynamic nature of service composition considered. Four detailed research works are planned. The first research work is the study of decomposing the user queries, which is in the form of nature language, and generating the service compositions automatically for the queries. The second research work is the study of refreshing the set of existing service compositions, with the usage records of the massive service compositions and individual services, as well as the dynamically evolutionary nature of service compositions taken into consideration. The third research work is the study of generalized modeling the context information, and how context-aware service composition recommendation can be achieved. The forth research work is the building of intelligent tourism service platform prototype system, which can be used to evaluate the theoretical results of this project.