中文摘要：

为更全面地考量服务质量（QoS）属性对服务最优选择的影响,进而提高服务组合的效率,该文提出了一种基于模糊层次分析法（FAHP）的多维QoS局部最优服务选择模型.模型首先定义了以QoS的真实度属性作为分量之一的多维QoS向量,并将QoS的真实度属性值作为赋权计算的依据,以使权重结果更合理、可信;之后,建立了包含双重质量属性（客观属性和主观评价）的模糊层次结构,以全面考量主客观QoS属性对服务最优选择的影响;在此基础上,构造了可根据决策者的偏好与侧重动态调整权重分辨率的FAHP赋权模式,该赋权模式与传统的赋权模式不同,该赋权模式下的权向量不是唯一的（但权重的排序唯一）,而是一个包含可调参数的函数,通过参数的调整,可保留（或剔除）决策者偏好（或放弃）的候选服务,一方面使服务使用者在选择服务时更理智,另一方面也促使服务提供者进一步完善服务质量;最后,利用组合权向量及序数效用函数值得到了局部最优服务.与文献中已有赋权模式的对比实验（包括候选服务效用值的对比、相关QoS属性值的对比等）及参数的敏感性分析说明了模型的有效性,其决策结果在一定程度上提高了局部最优服务选择的科学性及合理性.

英文摘要：

To consider QoS attribute＇s impact on optimal service selection more comprehensively and then improve the efficiency of service composition,this paper proposed a local optimal model of service selection of Muti-QoS based on Fuzzy Analytical Hierarchy Process（FAHP）.First,it defined the Muti-QoS vector,modelled the fidelity of QoS as one component of the Muti-QoS vector,and used the value of the fidelity of QoS as the basis of weighting to make the weight more reasonable and reliable.Then,it established a fuzzy hierarchy including dual attributes of QoS（objective attribute and subjective evaluation）to fully consider the subjective and objective QoS attribute＇s impact on optimal service selection.Furthermore,it developed a FAHP-based weighting mode,in which the resolution ratio of weight can be adjusted by the decision-maker＇s preference.Different from the traditional weighting mode,the proposed weighting mode in this paper makes the weight vector be non-unique（however,the order of the components is unique）and be a function that includes an adjustable parameter.By adjusting the relevant parameter,decision-maker can reserve（or eliminate）the corresponding candidate services according to his preference（or dislike）.In this way,it not only helps service user select service more reasonably,but also promotes service provider to further improve the service quality.Finally,the local optimal service was obtained through the values of combination weight vector and ordinary utility function.In addition,the sensitivity analysis of parameters was conducted to study the parameters＇ impact on optimal service selection.Through the comparison（the value of ordinary utility function,the value of QoS attribute,etc.）with the weighting modes of literature,experiment results indicated that the proposed model is feasible and efficient,and to some extent,the decision results improve the scientificity and rationality of the local optimal service selection.