基于QoS反向交叉预测的Web服务推荐系统研究

发布时间：2018-06-27 14:14

  本文选题：协同过滤 + 数据平滑机制　； 参考：《浙江大学》2013年硕士论文

【摘要】：随着云计算的发展,越来越多的应用以云端服务的形式开放,随之引发了Web服务数量的爆炸式增长,互联网上涌现出越来越多的功能相同但服务质量(QoS)不同的Web服务。面对如此庞大的服务集合,用户手工在服务注册中心或者搜索引擎上查找所需服务变得越来越困难。基于QoS的服务推荐,旨在从众多等功能服务中挑选出满足用户QoS需求的服务,已经成为服务计算(SOC)领域最炙手可热的研究方向之一。 以往的研究工作已经充分利用了候选服务的QoS属性来解决此问题,然而这些工作都是基于一个共同的前提,那就是假设所有候选Web服务针对目标用户的QoS值均已知。由于Web服务的QoS具有不确定性(同一个Web服务针对不同用户的QoS存在很大差异)以及不完整性(很少有用户曾调用过所有的候选Web服务),因而这种假设和实际应用的情况存在相当大的差距,换而言之,很多Web服务针对目标用户的QoS是未知的。 为解决QoS值缺失的问题,本文基于协同过滤的思想,提出一种创新的QoS预测算法DRaC。DRaC算法中引入了数据平滑机制,对训练集中的用户进行聚类操作,并利用各聚类中用户的历史QoS信息对预测系统的输入数据集进行数据平滑化预处理,可以有效提高系统的QoS预测准确度。不同于传统的基于协同过滤的预测算法,DRaC算法提出了反向交叉预测方法,充分合理利用了训练矩阵中相似度较低的用户与服务数据,可改善数据稀疏问题为预测系统带来的影响,优化了预测效果。此外DRaC算法提出基于用户反馈的信任度模型,在线统计学习用户对系统推荐结果的反馈信息,自动建立与维护用户信任度模型,并将其与QoS预测过程相结合,可以做到动态改善系统的预测效果。 最后,本文基于真实的QoS数据集验证了DRaC预测算法的效果,并通过实验分析了算法中各个参数对预测结果的影响。
[Abstract]:With the development of cloud computing, more and more applications are open in the form of cloud services, which has triggered an explosive increase in the number of Web services. More and more Web services with the same function but different quality of service (QoS) are emerging on the Internet. It is becoming more and more difficult to find the services needed. QoS based service recommendation, which is designed to select services that meet the user's QoS requirements from many other functional services, has become one of the hottest research directions in the service computing (SOC) field.
Previous research has made full use of the QoS attributes of candidate services to solve this problem. However, these work is based on a common premise that all candidate Web services are known for the QoS value of the target users. Because the QoS of the Web service is uncertain (the same Web service exists for the QoS of different users. There are great differences) and incompleteness (few users have ever called all candidate Web services), so there is a considerable gap between this hypothesis and the actual application, in other words, a lot of Web services are unknown to the target user's QoS.
In order to solve the problem of missing QoS value, based on the idea of collaborative filtering, this paper proposes an innovative QoS prediction algorithm DRaC.DRaC algorithm, which introduces a data smoothing mechanism to cluster the trained users, and uses the historical QoS information of the users in each cluster to preprocess the data set of the input data set of the prediction system. It can effectively improve the QoS prediction accuracy. Unlike the traditional collaborative filtering based prediction algorithm, the DRaC algorithm proposes a reverse cross prediction method, which makes full use of the low similarity of users and service data in the training matrix, and improves the impact of the data sparse problem to the prediction system and optimizes the prediction effect. In addition, the DRaC algorithm proposes a trust degree model based on user feedback. It can learn the feedback information of the user's recommendation results online, and automatically establish and maintain the user trust model, and combine it with the QoS prediction process, and can improve the prediction effect of the system dynamically.
Finally, based on the real QoS dataset, the effectiveness of the DRaC prediction algorithm is verified, and the influence of the parameters in the algorithm on the prediction results is analyzed through experiments.
【学位授予单位】：浙江大学
【学位级别】：硕士
【学位授予年份】：2013
【分类号】：TP393.09;TP391.3

【参考文献】

相关期刊论文 前3条

1 张成文;苏森;陈俊亮;;基于遗传算法的QoS感知的Web服务选择[J];计算机学报;2006年07期

2 李研;周明辉;李瑞超;曹东刚;梅宏;;一种考虑QoS数据可信性的服务选择方法[J];软件学报;2008年10期

3 陈彦萍;李增智;郭志胜;晋勤学;王创;;Web服务组合中基于服务质量的服务选择算法[J];西安交通大学学报;2006年08期

相关硕士学位论文 前1条

1 张慧;Web服务环境下单点登录与访问控制研究[D];中南大学;2008年

，

本文编号：2074182