云计算模式下Web服务QoS预测技术研究

发布时间：2018-02-11 22:54

本文关键词： Web服务 QoS预测服务推荐协同过滤粒子群算法　出处：《江西财经大学》2014年硕士论文　论文类型：学位论文

【摘要】：随着云计算和Web服务技术的快速发展,软件的开发方式也随着不断发生变化。传统的一次开发、持续使用为特征的软件开发模式正逐渐向软件即服务、按需提供、云端虚拟化这一新型模式转变。Web服务作为一种松耦合、复用性强、开放式的业务提供方式,已成为云计算中心中的一种重要服务资源,帮助用户构建软件业务流程和应用程序。近年来,部署在云中心或网络上的Web服务数量以惊人的速度增长,出现了大量功能等价的Web服务。在保证功能需求的前提下,现今人们更加强调Web服务的质量属性(Quality of Service, QoS)。因此,如何从众多功能等价的候选服务集合中选择出最适合用户QoS需求的服务便成为一个重要研究热点。应用预测技术来对Web服务的QoS进行预测是实现Web服务推荐的重要手段。本文主要对Web服务QoS数值预测和排序预测进行了研究。在Web服务QoS数值预测方面,提出了一种基于相似性感知的Slope One协同过滤算法SASO,该算法能够有效利用最近邻用户及用户本身的信息来实现对Web服务QoS的预测,并且算法中还利用了基于统计过程控制(Statistical Process Control, SPC)的数据抹平处理方法来处理数据集中的异常点。此外,为了验证SASO算法在QoS数值预测上的可行性和有效性,通过实验将SASO算法与现有的一些QoS预测算法进行了对比分析。结果表明,SASO算法在QoS数值预测上具备较好的效果。在Web服务QoS排序预测方面,本文提出了一种基于粒子群优化的排序预测算法PSORank,该算法通过寻找最近邻用户、构造适应度函数、优化搜索等手段来实现对Web服务QoS的排序预测。在预测过程中,该算法还充分考虑了服务优先级概率、用户群体调用行为等因素对预测的影响。通过实验对比分析,验证了PSORank算法在Web服务QoS排序预测上的可行性和有效性。结果表明,PSORank算法在Web服务QoS排序上具备较好的预测能力,并且能够较为有效地克服种群敏感性问题。通过上述研究,所形成的Web服务QoS预测方法对Web服务的有效推荐具备一定的参考价值,较为有效地提高用户对Web服务选择的满意度。
[Abstract]:With the rapid development of cloud computing and Web services technology, the way of software development is changing. Cloud virtualization is a new model transformation. Web services, as a loosely coupled, reusable and open mode of business provision, have become an important service resource in cloud computing center. To help users build software business processes and applications. In recent years, the number of Web services deployed on the cloud center or network has grown at an alarming rate, resulting in a large number of functionally equivalent Web services. Nowadays, more emphasis is placed on the quality of Web services. Therefore, how to select the most suitable services from the set of functional equivalent candidate services has become an important research hotspot. The application of prediction technology to predict QoS of Web services is an important means to realize the recommendation of Web services. This paper mainly studies the numerical prediction and ranking prediction of Web services QoS. In the aspect of Web service QoS numerical prediction, This paper presents a Slope One collaborative filtering algorithm based on similarity perception. The algorithm can effectively utilize the information of the nearest neighbor user and the user themselves to predict the QoS of Web services. In addition, in order to verify the feasibility and validity of the SASO algorithm in QoS numerical prediction, the statistical process control method based on Statistical Process Control (SPC-based) is used to deal with the outliers in the data set. The SASO algorithm is compared with some existing QoS prediction algorithms through experiments. The results show that the SASO algorithm has a good effect on QoS numerical prediction. In this paper, a sort prediction algorithm based on particle swarm optimization (PSO) is proposed. The algorithm realizes ranking prediction of Web service QoS by searching nearest neighbor user, constructing fitness function, optimizing search and so on. The algorithm also takes into account the influence of the probability of service priority and the behavior of user group call on the prediction. The feasibility and effectiveness of PSORank algorithm in Web service QoS ranking prediction are verified. The results show that the PSORank algorithm has better prediction ability in Web service QoS ranking, and can overcome the population sensitivity problem more effectively. The proposed QoS prediction method for Web services has a certain reference value for the effective recommendation of Web services and can effectively improve the satisfaction of users with Web service selection.
【学位授予单位】：江西财经大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TP393.09

【共引文献】