服务组合是快速构造增值服务应用的有效途径.现有的组合服务选择模型基于单个需求的假设,而在实际应用中,需求是大量的且在较小时间范围内认为是多个需求同时到达.这些需求在服务调用上的交集导致它们竞争或者共享使用某些服务,使得当前的服务组合方法无法有效应对此类场景.提出面向多需求的服务组合模型与算法.考虑到服务资源的独占及共享特性以及多个评价要素间的决定性优先关系,给出基于冲突避免调度和优先级别的评价方法.在此基础上针对遗传算法制定多需求权衡策略,提出面向多需求可满足性折中的服务组合方法.实验结果表明该方法可保证多个需求的可满足性之间的均衡,通过改进编码能够高效获得较优结果.与目前可能的其他策略相比,该方法对可用服务在数量、质量方面的不同情况表现出更优的适应性.
Service composition is effective in constructing value-added service rapidly for service-oriented applications.Existing selection models for composite services rely severely on assumptions that customers' each requirement is raised alone,while in reality,service requirements can be numerous in practical applications.And when a small time slide is focused on,multiple requirements can be seen as concurrent and service sets involved by sub-solutions corresponding to each individual requirement have intersections,resulting in competitions or sharing of certain services between requirements.Therefore,current single requirement-oriented methods can not deal with the situation that multiple service requirements arrive concurrently competing for services.This paper presents a multiple service requirements-oriented service composition model and algorithm.In the light that service can either be exclusive or sharable and decisive priority relations exist between all assessment factors,an assessment method based on confliction-avoidance scheduling and graded weighting priorities is put forward.On that basis,tradeoff strategies are proposed for genetic algorithm and a service composition method is put forward for tradeoff between satisfactions of multiple service requirements.Experiment results show that this method ensures proportionality of all sub-solutions and sub-optimal solutions can be gained efficiently by improving its coding manner.Compared with other possible strategies,it has proved superior applicability to different circumstances of quantity and quality of available services.