一种融合反向交叉预测的QoS预测方法  

A novel QoS prediction approach based on reversed and cross prediction

作　　者：洪朝群[1] 陈亮[2] 冯怡鹏[2] 吴健[2] 

机构地区：[1]厦门理工学院计算机学院,厦门福建361024 [2]浙江大学计算机学院,杭州310027

出　　处：《华中科技大学学报（自然科学版）》2013年第S2期137-142,共6页Journal of Huazhong University of Science and Technology(Natural Science Edition)

基　　金：国家自然科学基金资助项目(61202145);国家科技支撑计划资助项目(2011BAH16B04);国家高技术研究发展计划资助项目(2012AA02A604;2013AA01A604)

摘　　要：基于协同过滤的思想,提出一种QoS预测算法DRaC,该算法利用聚类技术实现QoS数据的平滑化预处理,同时利用反向预测和交叉预测机制解决数据稀疏问题,提高了预测准确度.最后基于真实的数据集,以NMAE值为标准,验证了DRaC算法相对于协同过滤等传统方法具有更好的预测效果,并衡量了各参数对预测结果的影响.A novel quality of service(QoS)prediction approach DRaC based on the idea of collaborative filtering was proposed.This approach first employed clustering technology to handle the data-smoothing pre-processing,and then employed reversed prediction and cross prediction to handle the problem of data sparsity for the purpose of improving the prediction accuracy.Experiment results of QoS prediction demonstrate that the proposed DRaC approach outperforms existing methods in terms of accuracy.Moreover,the impact of each parameter to DRaC was evaluated.

关 键 词：服务质量 协同过滤 反向预测 交叉预测 聚类技术 

分 类 号：TP393.09[自动化与计算机技术—计算机应用技术]