【作者】 侯宇；

【导师】 王成良；

【作者基本信息】 重庆大学 ， 计算机软件与理论， 2006， 硕士

【摘要】 Web服务是一种新型的Web应用程序,具有自包含、自描述、模块化和松耦合等特点,可以通过Web进行发布、查找和调用。简单地说,Web服务是“一次发布,到处可用”的新型Internet组件,支持服务复用和组合。服务发现,即帮助用户或某个Web服务找到最符合要求的服务,是当前Web服务的主要研究问题之一。目前,工业界推出了UDDI(Universal Description, Discovery, Integration)服务注册中心和服务自描述性的WSDL(Web Service Description Language)文件两个标准实现服务发现,但是两者组成的服务发现体系对基于功能的服务搜索不力,近来的研究多采用语义网的服务匹配方法解决这一新问题。这方面比较先进的是基于DAML-S(DARPA Agent Markup Language for Service,DARPA美国国防高级计划局)的服务匹配,这项技术通过DAML-S把服务定义成为服务本体,通过DAML+OIL(DARPA Agent Markup Language plus Ontology Interface Language)定义相应的Ontologies作为知识库,然后使用适当的推理机依据服务本体和相关联的知识库Ontologies进行逻辑推理,最后以某种匹配算法得到匹配的服务结果。这项技术同时具有高召回率和高精确率的优点,而且它完全采用XML格式,不仅方便互联网上传播,也有多种编程技术支持实现。本文在深入研究UDDI、WSDL和DAML-S的基础上,设计实现了“基于DAML-S的匹配器”。论文主要作了以下几个方面的研究:1.详细分析了UDDI和WSDL体系对服务发现的支持和不足之处,指出由于服务与网络文档本质的不同,并不适合采用关键字查询的方式来实现服务发现。Web服务的本质是互联网应用程序,直接表现服务功能的输入/输出参数是服务匹配的重点内容。2.分析了DAML-S的语法和结构以及应用于服务发现的优势,并在此基础上提出了匹配器的原理和构架。3.设计了基于DAML-S匹配器的匹配算法。算法主要分为两个部分:第一部分为需要语义支持的匹配算法,这部分需要定义知识库Ontologies实现语义支持,需要调用推理机实现匹配过程。对DAML-S的ServiceProfile结构设计了三个匹配算法,分别是输入参数、输出参数和Profile结构的匹配算法。第二部分是针对输入/输出参数的聚类匹配算法,这是不需要语义支持的基于统计的算法。补充这个算法是考虑到语义匹配的算法需要定义相应的知识库Ontologies才能执行,而在现实使用中,不可能总是在注册服务的同时就定义相应更多还原

【Abstract】 As web service is defined by most of researchers, it is a new type web application, which must have some characteristics, including self-subsumed, self-described, modularized and loosely coupled. Since web service can be published, found and called on line, it can be simply said that web service is one kind of new applicant modular on Internet, which could be published one time and be used everywhere. Also, this kind of modulus can be reused and can be combined to a more complex modular. As an applicant technique, web service is important to improve productivity and quality of software on Internet. But if web service satisfied these demands, there had been an automatic means to help user or another web service to find the most satisfying service and call it. This is one of main problems about web service, called service discovery.To implement service discovery, industry established two standards UDDI and WSDL. UDDI acts as a register center for services, and WSDL make the service could describe itself. But this couple of standards is not good enough, they don’t support capabilities searching for web service. By a new approach, service matching could resolve this problem.The emergence of the Semantic Web allows for the unambiguous interpretation of web resources and content through the use of shared web ontologies. Furthermore, ontologies based on DAML-S provide a semantic markup for Web Services. The services matching based on DAML-S, is excellent resolution for service matching. It describes the services as service ontologies by DAML-S at first, then according to the defined respective ontologies, as knowledge bases described by DAML+OIL , use an appropriate reasoner reasoning upon all ontologies, including service ontologies. At last, some algorithms send out result, a list of matched services. This approach provides high recall and precision at the same time. All used ontologies are based on XML pattern, which make them convenient to be transported on line and be implemeted by existing programming tools. Ambitious researcher of W3C, expect to combine web service with Semantic web to develop a newer framework, Semantic Service.In this article, it is analyzed how service discovery is supported by UDDI and WSDL. Also, it is analyzed the deficiencies about this architecture of service discovery. Because of the differents between web pages and web services, it is inappropriate to更多还原