Abstract: Web Service Discovery System（WSPS） storages and maintains the service information by a given topology system which is consisted with many service registers.The connection topology of the service registers can affect the scalability of the WSDS directly.Hierarchy model is introduced to defining dynamic tuple model in the data-level for resolving the isomer between diversified data source and data modal.The unified publish and query API customized to implement unified accessing mode in the abstract level.The scalability，agility and robustness of the WSDS is assured by a double hiberarchy topology structure.A distributed Web service discovery method has be designed and implemented.A two-phrase service location algorithm fitted double hiberarchy topology structure network is presented.Further，the time complexity，the space complexity and the messages complexity of the algorithms is analyzed.The simulation results show that，the Web service discovery method has obvious self-organized abilities and well scalability.It adapts the characteristics of Web service independence，dynamic changes，strong distribute.The two-phrase service location algorithm has high service location quality.

Key words: service registers, Web service discovery, dynamic tuple model, two-phrase service location algorithm

摘要： Web服务发现系统由一个或多个服务注册器连接组成一定的系统拓扑来存储和维护服务信息，服务注册器的连接拓扑直接关系到Web服务发现系统的可扩展性。采用层次模型在数据层定义统一的动态元组模型来解决各种数据源和数据模型的异构性，在抽象层定制统一的发布和查询API实现统一的访问方式，在网络层构造一种全新的双层结构保证整个系统的可扩展性、灵活性和鲁棒性，设计实现了一种分布式Web服务发现方法。给出了适应双层拓扑结构的两阶段查找算法，并对算法的时间、空间和消息复杂性进行了分析。实验结果表明，该Web服务发现方法具有明显的自组织特性和良好的可扩展性，适应Web服务自主，动态变化和强分布特点，同时两阶段查找算法具有较好的服务查找能力。

关键词: 服务注册器, Web服务发现, 动态元组模型, 两阶段查找算法