【目的】针对网络中海量的Web服务,提出一种有效的Web服务聚类与发现方法。【方法】利用BTM学习整个Web服务描述文档集的隐含主题,通过推理得出每个文档的主题分布,并进行聚类。在此基础上,创建一个快速的Web服务发现机制。【结果】与使用LDA和外部语料库等方法进行对比实验,本文方法的查准率和标准折损累计增益均有所提高。【局限】仅考虑服务的功能信息,未将服务的质量信息纳入算法。【结论】实验结果显示该方法可以更准确地发现符合用户需求的服务。 【Objective】 In view of the large number of Web services in the network, an effective Web service clustering and discovery method is proposed. 【Method】 Using BTM to learn the implicit topics of the entire Web service description document set, the theme distribution of each document is deduced and clustered. On this basis, create a fast Web service discovery mechanism. 【Result】 Compared with the method of using LDA and external corpus, the accuracy of the proposed method and the cumulative loss of standard impairment increased. [Limitations] Only consider the functional information of the service, the service quality information is not included in the algorithm. 【Conclusion】 The experimental results show that this method can find more accurate services that meet the needs of users.