Abstract: To deal with the Quality of Services(QoS)data fluctuation caused by random,fuzzy and inconsistent changes of Web services,a service selection approach considering the uncertainty of QoS data was proposed.Three eigenvalues of expectation,entropy and hyper entropy were used to predict QoS data distribution of Web service and to analyze the uncertainty of QoS data.Thus the Web services with wider QoS data fluctuation were rejected.On this basis,QoS attributes feedback similarities were calculated as the weights for QoS aggregation,and the service recommendation degree was computed according to QoS evaluations and the numbers of service users.A prototype system and a set of experiments were implemented to show that this approach could effectively reduce the influence of QoS data uncertainty on the service selection,thus the accuracy of service selection result could be improved,and the selected service would be more reliable.

Key words: Web service, service selection, quality of service, data uncertainty, attribute cloud

摘要： 针对Web服务随机性、模糊性和不一致性变化导致的QoS数据波动问题,提出一种考虑QoS数据不确定性的服务选取方法。该方法采用QoS属性云的期望、熵和超熵三个数字特征预测Web服务的QoS数据分布,分析QoS数据的不确定性,剔除QoS数据波动比较大的Web服务。在此基础上采用QoS属性反馈相似度作为计算QoS聚合值的权重,依据服务使用者的QoS评价及其数量计算服务的推荐度。通过一个原型系统和一组实验说明该方法能够有效减少QoS数据的不确定性对服务选取的影响,提高Web服务选取结果的准确性,使选取的服务更可靠。

关键词: Web服务, 服务选取, 服务质量, 数据不确定性, 属性云