基于依赖图模型的Web服务组合方法研究

发布时间：2018-05-15 09:46

本文选题：Web服务组合 + Web服务依赖图　；参考：《南京大学》2014年硕士论文

【摘要】：Web服务组合是一种能够通过组合多个功能简单的Web服务来完成一项复杂任务的有效方式。但面对当前大规模的Web服务,如何快速的组合出满足用户功能性需求和非功能性需求的Web服务组合结果是富有挑战性的问题。近几年来,Web服务组合问题逐步成为Web服务研究领域的热点,涌现出了很多解决Web服务组合问题的方法。基于图模型的方法是解决服务组合问题的主要方法之一,大量的研究者基于图模型对服务组合问题提供了有效的解决方案。本文基于依赖图模型对Web服务组合问题进行了初步研究,尝试从三个方面提出了进一步的解决方案。首先,对Web服务组合问题进行了介绍,针对Web服务组合过程中的搜索空间问题,提出了一种基于搜索空间约减的Web服务组合方法。该方法主要包含Web服务依赖图的建立过程、子节点产生过程和搜索树的搜索过程三个方面。重点介绍了Web服务依赖图的建立,分别分析了初始化过程和添加虚拟服务过程。提出了一种通过子节点产生树来生成子节点集的方法,大大减少了子节点产生数目,从而减小了搜索空间,提高了搜索效率。在多个数据集上的实验表明,该方法在服务组合结果搜索效率上有很好的提升效果。其次,针对大规模服务组合中QoS优化和组合结果中冗余服务的问题,我们提出了一种能够保持最优QoS属性值的同时约减冗余Web服务的Web服务组合方法。该方法主要包含QoS属性值的计算和约减冗余Web服务两个关键步骤。首先,基于Web服务依赖图,利用前向遍历的方式从前往后计算所有Web服务的QoS属性值,从而得到输出层的最优QoS属性值。然后通过约减冗余服务和反向遍历Web服务依赖图选取具有最优QoS属性值的Web服务集。在多个数据集上的实验表明,该方法能够在Web服务组合结果中保持最优QoS属性值的同时约减冗余Web服务。最后,针对动态网络环境下Web服务组合的适应性问题,提出了一种基于依赖图模型的服务组合结果修复的方法。该方法主要包含三个方面：首先,去除失效的Web服务及其后继服务；然后,考察该Web服务组合结果的可修复性；最后,重新计算QoS属性值和约减冗余服务得出次优QoS属性值的Web服务组合结果。在多个数据集上的实验表明,通过修复方式得出的Web服务组合结果要比通过重新组合方式得出的Web服务组合结果的效率高,具有较好的加速比。
[Abstract]:Web service composition is an effective way to accomplish a complex task by composing multiple simple Web services. However, in the face of the current large-scale Web services, it is a challenging problem how to quickly combine the composition results of Web services to meet the functional and non-functional requirements of users. In recent years, the Web services composition problem has gradually become a hot topic in the field of Web services research, and many methods to solve the Web services composition problem have emerged. Graph model-based approach is one of the main methods to solve the service composition problem. A large number of researchers provide an effective solution to the service composition problem based on the graph model. In this paper, based on dependency graph model, the Web service composition problem is preliminarily studied, and further solutions are proposed from three aspects. Firstly, the Web service composition problem is introduced, and a Web service composition method based on search space reduction is proposed to solve the search space problem in the process of Web service composition. The method mainly includes three aspects: the establishment of Web service dependency graph, the generation of child nodes and the search process of search tree. This paper mainly introduces the establishment of Web service dependency graph and analyzes the initialization process and the adding virtual service process respectively. In this paper, a method of generating the set of child nodes by generating trees of child nodes is proposed, which greatly reduces the number of child nodes, thus reduces the search space and improves the search efficiency. Experiments on multiple datasets show that the proposed method can improve the search efficiency of service composition results. Secondly, in order to solve the problem of QoS optimization in large-scale service composition and redundant services in composition results, we propose a Web service composition method which can maintain the optimal QoS attribute value and reduce redundant Web services at the same time. This method mainly includes two key steps: calculating the value of QoS attribute and reducing redundant Web service. Firstly, based on the Web service dependency graph, the forward traversal method is used to calculate the QoS attribute value of all Web services, and the optimal QoS attribute value of the output layer is obtained. Then the set of Web services with the optimal QoS attribute value is selected by reducing redundant services and traversing the Web service dependency graph. Experiments on multiple data sets show that the proposed method can maintain the optimal QoS attribute value in the Web service composition results while reducing the redundant Web services. Finally, aiming at the adaptability of Web service composition in dynamic network environment, a service composition result repair method based on dependency graph model is proposed. This method mainly includes three aspects: first, remove the invalid Web service and its successor service; then, investigate the repairable result of the Web service composition; finally, Recalculating the QoS attribute value and reducing the redundant service, the Web service composition results of the sub-optimal QoS attribute value are obtained. Experiments on multiple data sets show that the result of Web service composition obtained by repair is more efficient than that of Web service composition by re-composition, and has a better speedup.
【学位授予单位】：南京大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TP393.09

【相似文献】