中文摘要：

如何从大规模服务集合中快速而准确的发现目标服务是应用Web服务技术的关键．现有基于语义的Web服务发现方法应用实施难度大且效果存在较大提升空间．文中首先提出了Web服务注册的信息模型，该模型不受限于具体的Web服务模型和表达语言，支持接口语义标注和接口依赖关系的申明．进而提出了基于二分图匹配的语义Web服务发现方法，通过对二分图最佳匹配进行扩展，将服务匹配问题转化成二分图的扩展最佳匹配的求解问题，支持服务接口之间的依赖关系，从而提高服务发现的效果．一系列仿真实验表明文中方法不仅能较大提高服务发现的召回率和准确率，而且还能以线性时间复杂度满足用户请求．

英文摘要：

How to discover services for users efficiently is the key for the application of the Web service technology. The current semantic-based service discovery methods are of great inconvenience to be used and have much potential improvement for their performance. This paper proposes an information model of registered services that isn＇t limited to specific service models and languages and supports semantic-annotation and interface-dependency declaration. Then, a method of service discovery based on bipartite graph matching is proposed that translates the service matchmaking to the problem of extended optimal matching for bipartite graph. It enhances the recall rate and precision as it considers the interface dependency relations within a service in matchmak- ing. A serial of experiments illustrate the method not only improves its efficiency much, but also has a linear complexity in time consumption to fulfill users＇ requests.