文章摘要
谢苏,刘子巍,李克.基千特征加权ML-kNN的网页浏览业务KQI预测[J].高技术通讯(中文),2021,31(3):263~269
基千特征加权ML-kNN的网页浏览业务KQI预测
KQI prediction for web browsing based on feature weighted ML-kNN
  
DOI:10.3772/j.issn.1002-0470.2021.03.006
中文关键词: 特征选择;智能网络运维(AIOps);关键质量指标(KQI); k近邻(kNN);移动互联网(OTT);移动众包感知(MCS)
英文关键词: feature selection, AI for IT operations (AIOps), key quality indicator (KQI), k-nearest neigh­bor (kNN), over-the-top (OTI), mobile crowdsensing (MCS)
基金项目:
作者单位
谢苏  
刘子巍  
李克  
摘要点击次数: 933
全文下载次数: 612
中文摘要:
      传统以网络为中心的移动网络运维往往是在接到用户投诉时才采取相应补救措施,随着移动互联网(OTT)业务的高速发展,这一问题愈发突出。 如何在监测用户业务感知的 基础上对用户业务质量进行预测预警并及时于预,是提高移动业务保障能力和网络运维智能化水平的重要手段。 本文利用从普通用户终端上采集的海量业务感知数据,重点针对网页浏览业务,研究了ML-ReliefF算法在业务感知采样数据降维中的应用。 在此 基础上,将特征选择结果与多标记k近邻(kNN)算法相结合,提出了基于特征加权的多标记k近邻算法应用于业务关键质量指标(KQI)预测。 实验结果表明,该方法可有效提高 KQI预测质量。
英文摘要:
      Traditional network-centric mobile network operation often takes corresponding remedial measures when receiv-ing user complaints about service quality. With the rapid development of over-the-top (OTI) services, this prob-lem has become increasingly prominent. How to predict and warn the user's service quality and timely intervene based on the service perception monitoring is an important means to improve the intelligence of network operation. In this paper, the service perception data crowdsensed from massive user terminals are utilized, focusing on the web browsing service, and the ML-ReliefF algorithm in the dimension reduction of service perception data is applied. On this basis, combined with the feature selection results with the multi-label k-nearest neighbor (ML-kNN) algo-rithm, a feature weighted key quality indicator (ML-kNN for KQI) prediction is proposed. Experimental results show that this method can effectively improve the quality of key quality indicator (KQI) prediction.
查看全文   查看/发表评论  下载PDF阅读器
关闭

分享按钮