基于特征学习的5G用户投诉根因定位方法  

Root Cause Location Method of 5G User Complaints Based on Feature Learning

作　　者：岳烈骥 孙伟 侯普 吕程程 陈凯 Yue Lieji;Sun Wei;Hou Pu;Lu Chengcheng;Chen Kai(China Mobile Communications Group Beijing Co.,Ltd.,Beijing 100027,China;China Academy of Information and Communications Technology,Beijing 100191,China)

机构地区：[1]中国移动通信集团北京有限公司,北京100027 [2]中国信息通信研究院,北京100191

出　　处：《信息通信技术》2023年第3期75-79,共5页Information and communications Technologies

摘　　要：5G网络的复杂性和业务的多样性,使得5G用户投诉的根因定位涉及终端、锚点、5G站点等多种因素,导致用户投诉的根因定位困难,处理效率低。文章基于特征学习的5G用户投诉根因定位,使用数据分析和特征工程算法,基于大量5G用户的历史投诉记录,结合业务经验和统计学准则,筛选出5G投诉相关的关键指标集合。针对每位5G用户的投诉,根据用户的经纬度地址,匹配出投诉关联的5G基站,并对关联基站提取出指标数据集合,使用孤立森林算法,筛选出关联基站的异常指标集合,最终结合专家经验规则,得出用户投诉的根因。The complexity of the 5G network and the diversity of services make the root cause location of 5G user complaints involve multiple factors such as terminals,anchor points,and 5G sites,making it difficult to locate the root cause of user complaints and the processing efficiency is low.The root cause location of 5G user complaints based on feature learning,using data analysis and feature engineering algorithms,based on a large number of historical complaint records of 5G users,combined with business experience and statistical criteria,screens out a set of key indicators related to 5G complaints.For each 5G user's complaint,according to the user's latitude and longitude address,This paper matches match the 5G base station associated with the complaint,and extracts the index data set for the associated base station,uses the isolation forest algorithm to screen out the abnormal index set of the associated base station,and finally combines expert experience rules to find the root cause of user complaints.

关 键 词：5G 投诉根因定位 数据分析 特征工程 孤立森林 

分 类 号：TN929.5[电子电信—通信与信息系统]