基于大数据挖掘的移动互联网业务用户感知评价及应用研究

发布时间：2018-06-24 21:03

本文选题：移动互联网 + 用户感知　；参考：《北京邮电大学》2016年硕士论文

【摘要】：随着移动智能终端和移动网络的飞速发展,移动互联网业务也已由最初单一的短彩信业务发展到如今的移动游戏、移动社交、移动视频和移动即时通信等多种多样的新型业务。同时,移动网络运营商对网络质量关注的重点也从单一的语音业务转移到了种类繁多的移动互联网业务上。通常运营商对用户感知的评价是通过将网络质量KPI指标映射为用户感知指标来实现的,即用KPI指标中的接通率、掉话率、用户投诉等指标来衡量用户感知的好坏。但是,KPI指标与实际用户感知两者之间是有本质区别的,也就是说传统的KPI指标不能准确地反映用户的真实感受。用户感知的评判结果是否与用户实际感受相吻合,关键在于从网络侧指标中提取出用户使用业务的实际感知情况。本文从移动运营商网络侧丰富的数据源入手,基于大数据挖掘的结果,针对移动互联网主流业务,建立主流业务的用户感知评价体系,并探索基于用户感知的网络优化、流量经营、客户维系等方法,以保障移动运营商网络的精细化运营。主要内容为:首先,本文对移动网络的架构以及移动网络中所产生的数据源进行了详细描述,并结合数据源的特点对移动网络多数据源分析方法进行分析对比。同时,本文综述了移动互联网业务用户感知模型的研究现状,总结了目前用户感知模型研究中存在的问题。通过对不同的研究方法进行比对,衡量各种研究方法的优缺点,以此为基础加以改进产生了本文使用的评价方法。其次,本文重点介绍了基于大数据挖掘的移动互联网主流业务的用户感知建模方法。先对主流业务的评价指标进行选择,并应用系统聚类法对各类指标进行筛选、合并,最终确定模型搭建中使用的指标范围。之后,本文运用AHP层次分析法确定各类指标对各项业务的用户感知权重,最终形成基于单用户的主流业务用户感知评价模型。最后基于中国联通的现网数据,完成了某城市移动互联网主流业务的用户级及小区级的用户感知评价。最后,本文将所建立的用户感知模型与现网应用相结合,通过建立的用户感知模型来对现网中存在的问题进行分析筛查,并结合网络KPI,对问题进行精准定位,精准有效地解决现网中存在的网络问题,证明了本文所建立的用户感知模型的有效性及其在运营商工作中的重要意义。
[Abstract]:With the rapid development of mobile intelligent terminals and mobile networks, mobile Internet services have also developed from a single short MMS service to a variety of new services such as mobile games, mobile social, mobile video and mobile instant messaging. At the same time, mobile network operators focus on network quality from a single voice service to a wide variety of mobile Internet services. Usually the operators evaluate the user perception by mapping the network quality KPI index to the user perception index, that is, using the KPI index of the connection rate, drop rate, user complaints and other indicators to measure the good or bad of the user perception. However, there is an essential difference between the KPI index and the actual user perception, which means that the traditional KPI index can not accurately reflect the true feelings of the user. Whether the judgment result of user perception is consistent with the user's actual perception is the key to extract the actual perception of the user's using service from the network side index. This paper starts with the abundant data sources on the network side of the mobile operators, and based on the results of big data mining, establishes the user perception evaluation system for the mainstream services of the mobile Internet, and explores the network optimization based on the user perception. Traffic management, customer maintenance and other methods to ensure the fine operation of mobile operators network. The main contents are as follows: firstly, the architecture of mobile network and the data source generated in mobile network are described in detail, and the analysis methods of multiple data sources in mobile network are analyzed and compared according to the characteristics of data sources. At the same time, this paper summarizes the research status of mobile Internet service user perception model, and summarizes the existing problems in the research of user perception model. Through the comparison of different research methods, the advantages and disadvantages of various research methods are measured, and the evaluation methods used in this paper are improved on this basis. Secondly, this paper focuses on user aware modeling of mainstream mobile Internet services based on big data mining. Firstly, the evaluation index of mainstream business is selected, and the system clustering method is applied to screen and merge all kinds of indicators. Finally, the range of indicators used in the model building is determined. After that, this paper uses AHP to determine the user perceived weight of all kinds of indexes, and finally forms a single user based mainstream service user perception evaluation model. Finally, based on the current network data of China Unicom, the user perception evaluation of the mainstream mobile Internet service in a certain city is completed. Finally, this paper combines the established user perception model with the existing network application, through the established user perception model to analyze and screen the existing problems in the network, and combined with the network KPI to accurately locate the problems. It proves the validity of the user perception model established in this paper and its significance in the work of operators.
【学位授予单位】：北京邮电大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：TN929.5;TP393.0;TP311.13

【参考文献】