基于云模型的服务信誉度评估及其应用研究

发布时间：2018-04-21 22:37

本文选题：Web服务 + 服务选择　；参考：《辽宁大学》2014年硕士论文

【摘要】：Web服务作为面向服务体系架构(Service-Oriented Architecture, SOA)的一种实现，快速地应用于互联网体系结构中。然而数量激增的Web服务使得如何为用户选择满足需求的服务成为了棘手的问题。目前，针对服务选择问题的研究大多使用简单量化参数的方法评估服务信誉度，并以此为依据选择和发现服务。然而由于存在异常用户评价攻击和虚假QoS声明，服务信誉度评估不够真实准确，因此建立一个全面描述服务信誉度的模型非常重要。云模型能够全面的描述模糊概念，并能多角度反映实体的特点。云模型根据样本云滴的分布范围和规律，用云表达定性概念的特征，并通过云的参数Ex、En和He反映实体的真实水平和稳定程度，因而可以检测出异常用户的攻击和虚假的服务信息。本文将云模型运用到服务信誉度计算过程中，该模型可以全面的保留计算过程中参数的信息，避免单一数值表达的局限性，同时减小虚假评价和虚假QoS声明对计算服务信誉度的影响，使估更加准确。具体工作如下：首先，，建立用户评价主观服务信誉度评估模型。对云模型的输入样本进行分析，提出基于PeerTrust模型和用户历史评价信息的评价相似度算法，相似度作为云滴输入后，输出用户评价质量云，反映出用户评价的真实性和稳定性特征，剔除评价质量低于阈值的评价、惩罚行为波动的用户，使得服务信誉度计算结果更加真实准确，有效地抗击异常用户的虚假评价攻击。其次，建立服务提供者客观服务信誉度评估模型，即评估QoS声明的可靠性。统计云模型的输入云滴样本，提出关于QoS声明与实际值之间的相似度算法，并通过云模型输出服务的声明质量云，反映出QoS声明的真实性和稳定性，对于发布虚假QoS声明的服务，其信誉度将受到惩罚而降低。并将综合服务信誉度作为服务选择的依据。最后，将本文提出的基于云模型的服务信誉度计算方法与其他方法进行对比分析，在仿真实验环境中模拟了多个场景进行测试，结果表明，本文提出的方法在抗击虚假信息、有效识别波动行为和交易成功率上都有较好的表现，验证了本文提出方法的有效性。
[Abstract]:As an implementation of Service-Oriented Architecture (SOA), Web services are rapidly applied to Internet architecture. However, the proliferation of Web services makes it difficult for users to choose services that meet their needs. At present, most of the researches on service selection use simple quantitative parameters to evaluate the service reputation, and select and discover the service based on it. However, due to the existence of abnormal user evaluation attacks and false QoS statements, the evaluation of service reputation is not true and accurate, so it is very important to establish a comprehensive model to describe the service reputation. The cloud model can describe the fuzzy concept comprehensively, and can reflect the characteristics of the entity from many angles. According to the distribution range and law of sample cloud droplets, the cloud model expresses the characteristics of qualitative concept by cloud, and reflects the real level and stability of entity through the parameters of cloud E _ (n) and he. Therefore, abnormal user attacks and false service information can be detected. In this paper, the cloud model is applied to the service reputation calculation process. The model can keep the information of the parameters in the calculation process and avoid the limitation of the single numerical expression. At the same time, the influence of false evaluation and false QoS statement on the reputation of computing services is reduced, so that the estimation is more accurate. The specific work is as follows: First of all, the subjective service reputation evaluation model of user evaluation is established. After analyzing the input samples of cloud model, an evaluation similarity algorithm based on PeerTrust model and user history evaluation information is proposed. It reflects the authenticity and stability of the user evaluation, removes the evaluation of the evaluation quality below the threshold value, punishes the users whose behavior fluctuates, makes the calculation results of the service reputation more true and accurate, and effectively resists the false evaluation attack of the abnormal users. Secondly, an objective service reputation evaluation model for service providers is established to evaluate the reliability of QoS claims. The input cloud drop samples of the cloud model are counted, and the similarity algorithm between the QoS declaration and the actual value is proposed, and the declaration quality cloud of the service is outputted by the cloud model, which reflects the authenticity and stability of the QoS declaration. For services that issue false QoS claims, their reputation will be punished. And the comprehensive service reputation as the basis for service selection. Finally, this paper compares the cloud model-based service reputation calculation method with other methods, and simulates several scenarios to test in the simulation experimental environment. The results show that the method proposed in this paper is fighting against false information. It is proved that the proposed method is effective in identifying volatility behavior and trading success rate.
【学位授予单位】：辽宁大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TP393.09

【参考文献】