Research On Trustworthiness Evalaution Of Saas Service In Cloud Computing Environment

With the development of cloud computing technology,its application has become more and more extensive and various cloud services have become more and more important.However,cloud computing removes users' traditional control over data maintenance and brings new security issues.Compared with traditional security i.e.,hard security,trust as soft security is one of the most concerned key issues affecting the adoption and development of cloud computing.Trust evaluation has become an effective way to solve security problems.The Cloud Service Provider(CSP)and Cloud Service Consumer(CSC)need to trust each other in order to ensure the growth of cloud services.Ensuring reliable cloud service has become a hot topic in the field of cloud security research.In the basic cloud service delivery models,the SaaS service lies in the application layer.It is very significant to establish a relatively complete cloud service trustworthiness evaluation framework by analyzing deeply and combining the trust context from the perspective of SaaS consumers.It can not only effectively solve the trust problem between CSC and CSP,but also be very important for the popularization and promotion of SaaS applications.Based on this background,the research work have been done on the trustworthiness evaluation of SaaS cloud services.By deep investigation and analysis into the existing trustworthiness evaluation methods,models and frameworks,we have found that there exists the following three problems.First,there is a lack of combining the subjective preference of CSCs into the trustworthiness evaluation when based on the QoS attributes of cloud service.If the preference of CSCs are ignored,the evaluation results will not meet the personalized needs of CSCs.It will become meaningless.Second,the existing trustworthiness evaluation methods and models are mainly proposed based on users'feedback while often ignoring other information sources and lack of comprehensive consideration about the cloud service context especially from the perspective of sociology.Third,the trust evaluation methods based on CSPs' reputation usually lack the analysis of the correlation between multiple feature attributes.In addition,the existing research focus much more on the implementation of specific algorithms,lacking of consideration of universality and scalability.To solve the above problems,various subjective and objective factors that may influence the trust of SaaS consumers to their providers in cloud computing environment are investigated and analyzed deeply in this article.And further research on trust evaluation methods,models and frameworks are made,fully considering the characteristics of multidisciplinary cross trust concept itself.Three different dimensions of SaaS trustworthiness evaluation methods are proposed based on three different dimensions.A SaaS cloud service evaluation framework is proposed based on the above work and a typical model is presented.In detail,the main contributions of this paper are as follows:Firstly,a method of multiple attribute decision-making is put forward based on SaaS cloud service quality(QoS)and both subjective and objective aspects,namely the TOPSIS(Technique for Order Preference by Similarity to an Ideal Solution)trustworthiness comprehensive evaluation method.On the one hand,this method proposes to use the monitored value of QoS attribute rather than the feedback of users to conduct trust evaluation.At the same time,it proposes to adopt the objective entropy weight method for different QoS attributes in order to reduce the influence of false or artificial parameter information.Both above guarantee the objectivity of QoS evaluation.On the other hand,this method also introduces the concept of trust preference which reflects the subjectivity of CSC consumers' trust.This will provide them with greater satisfaction.By organically combining the above two weights and applying them to the TOPSIS method,a novel QoS trustworthiness evaluation scheme of SaaS service is finally formed.Based on the evaluation tool made by the author and existing QWS 1.0 dataset in the real Web service,two groups of experiments and result analysis are carried out,which prove the feasibility and effectiveness of this method,as well as better user satisfaction.Secondly,a method to predict and evaluate the trustworthiness of CSC's unknown cloud services by using the social network of SaaS consumers is proposed.First,this method constructs the social trust network of CSCs based on the direct trust relationship in their social network and mutual trust relationship.Second,depending on the trust context of CSCs and the social influence of related principles,this paper proposes a method of using local social influence factors to calculate trust and applies it into the improvement of CF(Collaborative Filtering)recommendation algorithm to form a new social trust recommendation algorithm combined with social trust network and the local social influence.It can predict the trust degree for CSCs about the unknown cloud services.The method can improve the prediction accuracy of the traditional CF algorithm and alleviate the cold start problem of new users to some extent.The experimental results based on Epinions dataset from the real social networks show that the proposed method is more accurate in prediction than other three evaluation methods based on CF.Meanwhile,the improved CF recommendation algorithm can decrease the appearance of extreme prediction values to some extent by using Pearson Correlation Coefficient(PCC).Thirdly,a method based on Logistic Regression is proposed to classify and evaluate the trustworthiness of SaaS service or its provider.Based on the idea of the credit scoring card model in economics,this method firstly divides the feature attributes into boxes based on WoE(Weight of Evidence)transformation,then calculates the ?(Information Value)value and makes feature selection,and finally uses Logistic regression algorithm to conduct classification and evaluation.It can satisfy the rough classification needs of CSCs in evaluating SaaS service trustworthiness.Meanwhile,it can overcome the weakness caused by excessive number of features.According to the real dataset QWS1.0,the experiment is conducted to compare different types of classification and evaluation methods based on machine learning.The results show that the proposed method is superior to the other five classification algorithms in prediction performance of trustworthiness classification.Fourthly,based on the previous work,a comprehensive trustworthiness evaluation framework of SaaS cloud service with three different dimensions is proposed.These three dimensions include QoS attribute of SaaS cloud service,social trust network of CSCs and reputation evaluation of CSPs.The trustworthiness evaluation of these three dimensions can be carried out independently,or combined organically as a whole according to actual needs.That is,each of them can be assigned a certain weight respectively and then aggregation is carried out on the basis of different evaluation results obtained from each dimension.Finally,a model is proposed based on the framework.To sum up,this paper mainly proposes three kinds of SaaS cloud service trustworthiness evaluation methods based on different dimensions,and each method is verified by experiments respectively.On the basis of the above work,a comprehensive framework with three dimensions for SaaS service trustworthiness evaluation and a related model are proposed.