【作者】 罗叶飞；

【导师】 刘建勋；

【作者基本信息】 湖南科技大学 ， 计算机科学与技术， 2011， 硕士

【摘要】 面向服务的计算(SOC)和面向服务的架构(SOA)最近成为了产业界和学术界关注的焦点。在SOA和SOC中,Web服务扮演了关键与核心的角色。通过重用或合成Web服务来构造的新的服务或软件作为一种全新的软件开发模式吸引了广泛的兴趣,并产生了深远的影响。然而,随着Web服务数量的迅速增长,如何根据用户的需求,快速、准确发现Web服务成为了制约Web合成与应用的瓶颈。传统的基于UDDI的服务发现技术具有查询过程复杂、查全率和查准率都不高等缺点,为此研究人员提出了许多改进服务发现的方法,如基于各种语义Web服务模型的方法。然而,迄今为止,现有的服务发现技术通常将Web服务看成是孤立的,只考虑服务本身的属性,而很少考虑和利用服务之间的关联关系。如何挖掘和利用服务之间的关联关系来提高服务发现的效率和质量尚未得到足够重视。本文研究了如何挖掘Web服务之间的潜在关联关系挖掘并将其应用于服务发现。基于服务之间的关联关系,可以构造服务网络。利用服务网络和服务关系,用户可以像使用Web一样浏览服务和在服务间导航,在用户浏览服务时快速聚集相关服务,为用户下一步决策提供支持,大大降低了服务发现的使用门槛。具体地,本文的主要工作和贡献如下:(1)为了发现Web服务之间的相似关系,提出了基于TF-IDF的Web服务相似关系挖掘算法。其原理是对WSDL文档内容进行分析,构造特征向量,通过计算特征向量之间的相似性来计算相应服务之间的相似性。(2)为了发现服务操作之间潜在的可组合关系,提出了针对Web服务操作的组合关系挖掘算法。该算法的步骤主要包括:从WSDL文档中提取操作信息、基于WordNet计算输入/输出操作的名字和参数的相似性、记录匹配的操作对等。(3)使用真实的Web服务数据,我们对基于操作级组合关系的Web服务图结构进行了分析。(4)开发了相关工具和系统原型。提出将服务关系应用于我们原有的Web服务超市系统中,为用户提供相似服务发现和推荐功能,便于在服务间导航。测试结果验证了本文算法的有效性。更多还原

【Abstract】 Service-Oriented Computing (SOC) and Service-Oriented Architecture (SOA) has recently become the attention of the industry and academia. In the SOA and SOC, Web services play a crucial and central role. Reusing or synthesis Web services to construct the new Web services or software has attracted wide spread interest as a new software development model, and it’s got profound impact. However, with rapid growth in the number of the Web services, how to quickly and accurately find Web services base satisfy the user’s needs become the bottle neck to restricted Web services Synthesis and Application. Traditional UDDI-based service discovery technology has the disadvantages of query complexity, recall and precision, for which researchers have proposed many ways to improve service discovery, such as Semantic Web services model based on a variety of ways. However, recently, the existing service discovery technologies usually take the Web service as isolated, only consider the properties of the service itself, with little consideration and relationship between the uses of Web services. How to mining and use the services relationship to improve the efficiency and quality of service that has not been enough attention.This paper studies how to tap the potential Web services relationship between the mining and applied to service discovery. Relationship between service-based can construct service networks. Through using the services networks and the services relationship, users can browse the same as using Web services and navigation between services, browsing services in the user gathered related services quickly, making the next step for the user to provide support services and greatly lower the Web service discovery. Specifically, the main work and contribution of this paper is as follows.(1) In order to find similarities between the relationships between Web services, this paper proposed the algorithm of Web-based TF-IDF service similar relationship mining. The principle is, firstly, analysis the WSDL document content, and then structural feature vector and calculating the similarity between feature vectors to calculate the similarity between the corresponding services.(2) In order to find a service operation can be combined between the potential relationships, this paper proposed relationship between Web services composition algorithm for mining operations. The steps of the algorithm include: extraction operation from the WSDL document information, and then calculation of input/output operation sand parameters of the similarity of the name based on WordNet, then record matching operation on the other.(3) We use real Web services data, analyzed Web services composition graph structure which is based of relationships based on operation-level.(4) The development of tools and system prototypes. We proposed service relationship should be applied to our original Web service supermarket system, to provide users with similar service discovery and recommendation features, and easy to navigation between services for users. The last test results also verify the effectiveness of this paper’s algorithm.更多还原