服务双方偏好共识下的模糊天际线服务组合

对一些复杂应用,有时需要从众多Web服务中选出符合需求的原子服务并组合成增值的服务.随着大量Web服务的涌现,如何选择合适的原子服务变得十分困难.提出一种模糊天际线服务组合方法,该方法综合考虑用户的功能需求,以及用户和服务提供者对QoS的偏好,能够实现服务双方的利益最大化.首先,提出一个模糊Web服务选择框架;然后,通过引入一个服务属性偏好的共识方法,将用户和服务提供者的偏好关系转换成数值权重,并用于计算组合方案的切比雪夫距离;接着,通过一个演化算法(DWTD-EA)从组合方案集中寻找满意的天际线组合方案.最后,通过实验对所提出的算法和模型的有效性和鲁棒性进行验证.从实验可以看出:在top-k最优偏好的天际线方案验证实验中,当演化到60代以后,超过85％的优秀方案都能被发现;从DWTD-EA3种交叉(即单点交叉、两点交叉和均匀交叉)的收敛行为看出,当演化到55代以后,单点交叉能收敛到80％,而两点交叉和均匀交叉都能收敛到85％以上.

Fuzzy Skyline Web Services Composition with a Consensus between Service Providers and Users

For some complicated applications,Web service composition is a key issue to create value-added composite Web services from existent services,provided by different service providers.However,with an increasing number of Web services,selecting the ideal services for each task is difficult.To meet this challenge,a fuzzy Web services selection model was presented,then a consensus method was proposed for transforming the preference relations of users and providers into numeric weights that were used in the Tchebycheff distance computing. Finally an evolutionary algorithm was used to find efficiently preferred skyline solutions. Through a set of experiments, the effectiveness and robustness of this approach were evaluated.For the convergence results of top-k optimization objective, after having evolved about 60 generations,greater than 85% of optimal solutions could be found.For three crossover types of DWTD-EA, i.e. single-point,two-point and uniform crossover, when it evolved above 55 generations, single-point could converge to 80%, and the rest two types all could converge above 85%.