基于QoS及协同过滤的Web服务推荐方法研究
发布时间：2018-11-03 11:20
【摘要】：随着互联网技术的不断发展，Web服务推荐与选择已经逐渐成为学术界和工业界共同关注的重要研究内容,服务质量(QoS)是成功进行Web服务推荐的关键性因素。然而，，Web服务的QoS值在运行时刻可能会因为服务器超载，网络条件等多种因素的影响而发生变化。由于Web服务环境的动态性，目前现有的服务选择方法通常无法有效地涵盖QoS内在的不确定性，使得服务推荐结果与实际结果偏差较大。为解决Web服务的QoS值的动态性以及目前算法忽视QoS内在的不确定性,导致服务选择可靠性差问题，本文提出一种改进的基于协同过滤的Web服务推荐方法，该方法的引入使得服务用户不需要对Web服务进行调用，只需要对历史的Web服务的QoS信息进行分析挖掘就能找出适合用户的最优Web服务。 本文提出的推荐算法不同于传统的推荐算法，主要表现在以下几个方面：在服务可靠性方面，本文引入云模型中的逆向云算法来解决QoS内在不确定性导致的服务选择可靠性差问题，把不可靠的服务剔除；在相似度计算方面，本文算法在计算用户间相似度时，充分考虑了Web服务的内在特征，在计算服务间相似度时，充分考虑了用户的内在特征；在对QoS缺省值预测方面，为了缓解负数对预测性能的影响，本文对传统的基于服务的QoS预测算法和基于用户的QoS预测算法进行改进；当为目标用户预测的QoS值为负数时，使用服务或者用户QoS值算术平均的方法进行计算填充。最后联合基于服务的QoS预测算法和基于用户的QoS预测算法采用自适应均衡权重的方法给出最终的QoS预测结果。为验证本文提出算法的优越性，本文使用了真实环境下大规模的QoS数据集进行仿真实验，该数据集包含了1500000条Web服务调用记录，通过仿真对比实验证明了本文算法的优越性。
[Abstract]:With the continuous development of Internet technology, Web service recommendation and selection has gradually become an important research content of academia and industry. Quality of service (QoS) is the key factor for successful Web service recommendation. However, the QoS value of Web services may change at runtime due to the influence of server overload, network conditions and other factors. Because of the dynamic nature of the Web service environment, the existing service selection methods usually can not effectively cover the inherent uncertainty of QoS, which makes the service recommendation results deviate greatly from the actual results. In order to solve the dynamic QoS value of Web services and ignore the inherent uncertainty of QoS in current algorithms, this paper proposes an improved Web service recommendation method based on collaborative filtering, which results in poor reliability of service selection. With the introduction of this method, service users do not need to invoke Web services, but only need to analyze and mine the QoS information of historical Web services to find out the best Web services suitable for users. The recommendation algorithm proposed in this paper is different from the traditional recommendation algorithm, mainly in the following aspects: in terms of service reliability, In this paper, the reverse cloud algorithm in cloud model is introduced to solve the problem of poor reliability of service selection caused by the inherent uncertainty of QoS, and the unreliable services are eliminated. In the aspect of similarity calculation, when computing the similarity between users, the algorithm takes into account the inherent features of Web services, and the inherent characteristics of users when computing the similarity between services. In the aspect of QoS default prediction, in order to mitigate the influence of negative number on prediction performance, this paper improves the traditional QoS prediction algorithm based on service and the QoS prediction algorithm based on user. When the predicted QoS value for the target user is negative, the service or the user QoS arithmetic average method is used to calculate the population. Finally, the QoS prediction algorithm based on services and the QoS prediction algorithm based on users are combined to give the final QoS prediction results using the adaptive equalization weight method. In order to verify the superiority of the proposed algorithm, this paper uses a large scale QoS data set in real environment to carry out simulation experiments. The dataset contains 1500000 records of Web service calls, and the superiority of the proposed algorithm is proved by simulation and comparison experiments.
【学位授予单位】：南京邮电大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TP393.09
[Abstract]:With the continuous development of Internet technology, Web service recommendation and selection has gradually become an important research content of academia and industry. Quality of service (QoS) is the key factor for successful Web service recommendation. However, the QoS value of Web services may change at runtime due to the influence of server overload, network conditions and other factors. Because of the dynamic nature of the Web service environment, the existing service selection methods usually can not effectively cover the inherent uncertainty of QoS, which makes the service recommendation results deviate greatly from the actual results. In order to solve the dynamic QoS value of Web services and ignore the inherent uncertainty of QoS in current algorithms, this paper proposes an improved Web service recommendation method based on collaborative filtering, which results in poor reliability of service selection. With the introduction of this method, service users do not need to invoke Web services, but only need to analyze and mine the QoS information of historical Web services to find out the best Web services suitable for users. The recommendation algorithm proposed in this paper is different from the traditional recommendation algorithm, mainly in the following aspects: in terms of service reliability, In this paper, the reverse cloud algorithm in cloud model is introduced to solve the problem of poor reliability of service selection caused by the inherent uncertainty of QoS, and the unreliable services are eliminated. In the aspect of similarity calculation, when computing the similarity between users, the algorithm takes into account the inherent features of Web services, and the inherent characteristics of users when computing the similarity between services. In the aspect of QoS default prediction, in order to mitigate the influence of negative number on prediction performance, this paper improves the traditional QoS prediction algorithm based on service and the QoS prediction algorithm based on user. When the predicted QoS value for the target user is negative, the service or the user QoS arithmetic average method is used to calculate the population. Finally, the QoS prediction algorithm based on services and the QoS prediction algorithm based on users are combined to give the final QoS prediction results using the adaptive equalization weight method. In order to verify the superiority of the proposed algorithm, this paper uses a large scale QoS data set in real environment to carry out simulation experiments. The dataset contains 1500000 records of Web service calls, and the superiority of the proposed algorithm is proved by simulation and comparison experiments.
【学位授予单位】：南京邮电大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TP393.09
【参考文献】
相关期刊论文 前8条
1 李德毅，孟海军，史雪梅;隶属云和隶属云发生器[J];计算机研究与发展;1995年06期
2 冯在文;何克清;李兵;龚平;何扬帆;刘玮;;一种基于情境推理的语义Web服务发现方法[J];计算机学报;2008年08期
3 邓水光;尹建伟;李莹;吴健;吴朝晖;;基于二分图匹配的语义Web服务发现方法[J];计算机学报;2008年08期
4 张富国;徐升华;;推荐系统安全问题及技术研究综述[J];计算机应用研究;2008年03期
5 邓爱林,朱扬勇,施伯乐;基于项目评分预测的协同过滤推荐算法[J];软件学报;2003年09期
6 岳昆,王晓玲,周傲英;Web服务核心支撑技术:研究综述[J];软件学报;2004年03期
7 邵凌霜;周立;赵俊峰;谢冰;梅宏;;一种Web Service的服务质量预测方法[J];软件学报;2009年08期
8 刘建国;周涛;汪秉宏;;个性化推荐系统的研究进展[J];自然科学进展;2009年01期
相关博士学位论文 前1条
1 高e
本文编号：2307634
本文链接：https://www.wllwen.com/guanlilunwen/ydhl/2307634.html
