基于PRWDR算法的用户网络服务推荐方法及其实验结果分析  

User Network Service Recommendation Method and Experimental Results Analysis Based on PRWDR Algorithm

作　　者：刘微雨[1] LIU Wei-yu(Shaanxi Xueqian Normal University,Xi'an 710100 China)

机构地区：[1]陕西学前师范学院,陕西西安710100

出　　处：《自动化技术与应用》2020年第9期48-51,共4页Techniques of Automation and Applications

摘　　要：以相似邻居预测候选服务QOS值作为参考依据,选择部分最优QOS值组成的服务作为节点,连接不同服务的相似能相得到连边,提出了以随机游走以及多样性图排序作为分析依据的个性化服务推荐方法。根据实验结果来判断与PRWDR相关的二个参数λ、θ对算法性能产生的影响。确定最优的阈值θ=0.5和λ=0.7。对PRWDR对推荐准确性与功能多样性进行了综合分析,从而可以达到最优的表现效果。采用PRWDR算法得到的推荐结果形成更加均匀的分布结果,50部经典电影产生的次数在总数中只占到了52.2%,更能符合用户多样化需求。Taking the QOS value of candidate services predicted by similar neighbors as the reference basis,the service composed of some optimal QOS values is selected as the node,and the similarity energy of different services is connected to get the edge.A personalized service recommendation method based on random walk and diversity graph sorting is proposed for analysis.Evaluate the influence of two parameters of PRWDR the on algorithm performance is judged based on experimental results.Determine the optimal threshold forθ=0.5 orλ=0.7.The accuracy and functional diversity of recommendation by PRWDR are comprehensively analyzed,so as to achieve the optimal performance effect.The recommendation results obtained by PRWDR algorithm form a more uniform distribution result,and the number of 50 classic movies only accounts for 52.2%of the total number,which is more in line with the diversified needs of users.

关 键 词：服务推荐 数据稀疏性 多样性 随机游走模型 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]