摘要
全局最优和局部最优是服务选择的两种策略.现有的全局最优服务选择算法提供端对端约束下最优单解而非可接受的多解,既无法充分体现用户偏好和服务个性,也不利于激励服务提供者优化服务质量.首先,在引入序数效用函数作为局部服务排序的数值尺度的基础上,提出一种基于多维服务质量的局部最优服务选择模型MLOMSS(Multi-QoS based Local Opti mal Model of Service Selection),为自动选取优质服务提供重要依据.然后,构造客观赋权模式、主观赋权模式和主客观赋权模式来确定各服务质量属性的权重,既体现用户偏好和服务质量的客观性,又有助于快速生成聚合服务链.最后,通过语义Web服务集成平台SEWSIP(Semantic Enable Web Serv-ice Integration Platform)证明MLOMSS模型的有效性和灵活性.
Global optimal and local optimal are two strategies of service selection.The current global optimal algorithms of service selection provide single optimal solution instead of multi-acceptable solutions under end-to-end constraints,which cannot fully reflect users＇ preference and service personality,and is not conducive to encourage service provider to optimize the service quality.In this paper,an ordinary utility function is used as a numerical scale of ordering local services,and meanwhile a Multi-QoS based Local Optimal Model of Service Selection（MLOMSS）is proposed firstly to provide important grounds to choose the best service.Then,subjective weight mode,objective weight mode,and subject-objective weight mode are constructed to determine the weight of each QoS property,which not only shows users＇ preference and the objectivity of service quality but also helps to generate service composition chains.At last,experimental results indicate the flexibility and effectiveness of this model based on SEWSIP（Semantic Enable Web Service Integration Platform）.
出处
《计算机学报》
EI
CSCD
北大核心
2010年第3期526-534,共9页
Chinese Journal of Computers
基金
国家"九七三"重点基础研究发展规划项目基金(2007CB310803)
中国博士后科学基金(20070410061)
武汉大学软件工程国家重点实验室开放基金项目(SKLSE20080704)资助~~
关键词
局部最优服务选择模型
序数效用函数
主客观赋权模式
local optimal model of service selection
ordinary utility function
subject-objective weight mode
