Web服务可信性评估与推荐方法研究

发布时间：2019-03-16 14:18

【摘要】：Web服务组合具有功能丰富、跨平台、便利等优点,由于其所具有的优点,Web服务组合正在改变传统的软件开发模式。随着Web服务组合的兴起,Web上发布的服务种类越来越丰富,服务的数量也越来越庞大,大量的Web服务一方面给人们提供了更多的选择,另一方面,也使人们在进行服务组合的时候面临着无从选择的问题,人们必须要考虑服务是否安全、是否适合等问题。因此,一个有效的服务推荐系统是必不可少的,它能在用户面临服务选择的时候向用户推荐适合的服务。在此种情况下,本文对Web服务的可信性评估及其推荐方法进行了研究,主要目的是为了寻找一种可行的服务推荐方法,向用户推荐适合的服务。在本文的研究中,主要对服务可信属性等级评估、可信服务推荐方法和可信服务推荐中属性冲突问题进行了研究。在服务可信属性等级评估的研究方面,主要对前人的研究工作进行了分析和总结,在此基础上将前人的研究方法进行了融合和改进使其适合对服务可信属性的等级评估。在此方面,本文的主要贡献有：1、构造了可信属性等级评估树。2、将层次分析法引入到对单个可信属性的评估中。3、在综合多个专家计算得到节点可信值时考虑了每个专家的权重。在对服务可信属性等级评估进行研究之后接着对服务推荐方法进行了研究,对服务可信属性等级的评估主要是为了给服务推荐方法提供数据。在服务推荐方法的研究中,由于电子商务推荐与服务推荐有一定的相似性,首先对电子商务推荐和服务推荐进行了研究,总结了两者的异同之处,抓住电子商务推荐与服务推荐在评分上的差异：电子商务中用户对于商品的评分是唯一的,而在服务推荐中,用户对于服务的评分会随着服务的使用环境而改变。同时,在服务推荐中,服务组合后服务各可信属性之间并不是简单的叠加关系,有些属性之间会相互促进如安全性与可靠性之间,而有些属性之间会相互消弱如安全性与易用性之间,因此,为了使推荐的服务能够在组合后各个可信属性等级满足用户对组合应用可信等级的需求,需要解决组合过程中属性冲突的问题。基于以上所诉本文提出了一种服务推荐方法,该方法属于协同过滤推荐方法。在该部分本文主要的贡献为：1、分析总结了电子商务推荐与服务推荐的异同之处。2、提出了一种可信的服务推荐方法。3、解决了服务推荐过程中属性冲突的问题。最后,基于以上提出的服务推荐方法,本文给出了一个服务推荐案例,并对提出的推荐方法进行了模拟实验和分析,实验证明,本文提出的方法能够满足用户对于可信服务选择的需求。
[Abstract]:Web service composition has the advantages of rich function, cross-platform, convenience and so on. Because of its advantages, Web service composition is changing the traditional software development mode. With the rise of Web service composition, the types of services published on Web become more and more abundant, and the number of services becomes more and more huge. On the one hand, a large number of Web services provide people with more choices, on the other hand, People also have to consider whether the service is safe or not and whether it is suitable and so on, which makes people face the unselectable problem when they make the service composition. Therefore, an effective service recommendation system is essential. It can recommend suitable service to users when they are faced with service choice. In this case, this paper studies the credibility evaluation and recommendation methods of Web services. The main purpose is to find a feasible service recommendation method and recommend suitable services to users. In the research of this paper, the evaluation of service trusted attribute level, the method of trusted service recommendation and the problem of attribute conflict in trusted service recommendation are mainly studied. In the aspect of service trusted attribute evaluation, the previous research work is analyzed and summarized. On this basis, the former research methods are merged and improved to make it suitable for the evaluation of service trusted attribute level. In this respect, the main contributions of this paper are as follows: (1) A trust attribute rating evaluation tree is constructed. 2. The analytic hierarchy process (AHP) is introduced into the evaluation of a single trusted attribute. The weight of each expert is considered when the creditable value of nodes is obtained by synthesizing multiple experts. After the research on the evaluation of the service trust attribute level, the service recommendation method is studied. The evaluation of the service trust attribute level is mainly to provide data for the service recommendation method. In the research of service recommendation method, because of the similarity between e-commerce recommendation and service recommendation, this paper firstly studies E-commerce recommendation and service recommendation, and summarizes the similarities and differences between them. Take advantage of the difference between e-commerce recommendation and service recommendation in scoring: in e-commerce, the user's score for goods is unique, and in service recommendation, the user's score for service will change with the use of the service environment. At the same time, in the service recommendation, there is not a simple overlay relationship between the trusted attributes of the service after the service composition, and some attributes promote each other, such as security and reliability. And some attributes will weaken each other, such as security and ease of use. Therefore, in order to make the recommended service satisfy the user's requirement of the composition application trust level after the composition, each trusted attribute level can satisfy the user's need for the composition application trust level. The problem of attribute conflict in combination process needs to be solved. Based on the above, this paper proposes a service recommendation method, which belongs to collaborative filtering recommendation method. In this part, the main contributions of this paper are as follows: 1, the similarities and differences between E-commerce recommendation and service recommendation are analyzed and summarized. 2. A trusted service recommendation method is proposed. 3, which solves the problem of attribute conflict in the process of service recommendation. Finally, based on the above proposed service recommendation method, this paper gives a service recommendation case, and carries on the simulation experiment and the analysis to the proposed recommendation method, the experiment proves, The method proposed in this paper can meet the needs of users for the choice of trusted services.
【学位授予单位】：云南大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TP393.09

【参考文献】