【作者】 李威；

【导师】 廖青；

【作者基本信息】 北京邮电大学 ， 电子与通信工程（专业学位）， 2013， 硕士

【摘要】 随着移动互联网和嵌入式终端技术的快速发展和普及,移动互联网已经成为人们访问网络的重要途径之一,通过用户的行为分析获得用户的需求是电信运营商抢占市场份额,提高业务增长点的重要切入点。移动互联网用户行为分析从广义上涉及移动互联网用户的所有行为活动,包括网络心里活动、网络社会活动和网络访问活动；从狭义上讲主要涉及用户访问网络的行为规律。电信运营商通过获取用户访问移动互联网的行为规律,能够有效地配置网络资源并提供具有针对性的网络服务。因此,建立一套综合的移动互联网用户行为分析方法及提供有效的用户行为分析方案已经成为电信运营商在移动互联网业务拓展中亟待解决的问题。本文主要针对提供移动互联网用户行为分析的方法及可行的分析方案展开研究,在对移动互联网用户行为分析关键技术分析的基础上,结合数据挖掘技术、文本分类技术及样本约简与均衡技术,提出了一种基于KNN的样本约简及密度均衡分类算法,并在算法的基础上提出了移动互联网用户行为分析方法及实验方案。本文工作主要体现在：(1)根据移动互联网用户兴趣习惯的分析需求,设计了一种基于KNN的样本约简及密度均衡分类算法。该算法在满足用户行为分析的基础上,实现了对样本空间的约简和密度均衡,从效率和正确率两方面提高了移动互联网中用户兴趣习惯分析的能力。(2)针对移动互联网用户行为分析,提出了移动互联网用户行为分析方法与步骤,为移动互联网用户行为分析提供了方法基础。结合移动互联网用户行为分析方法、步骤及提出的基于KNN的样本约简及.密度均衡分类算法设计了移动互联网用户行为分析方案,并给出了具体实验。从移动互联网用户访问网络时间、网络流量及用户兴趣习惯三个维度实现了对用户行为的分析,实验结果表明所提的方法、算法及方案在移动互联网用户行为分析过程中是有效可行的。更多还原

【Abstract】 With the rapid development and popularization of the mobile Internet and embedded terminal technology, mobile Internet has become an important way for people to access the network, user behavior analysis is the importance of telecom operators to seize market share and business growth entry point. Mobile Internet user behavior analysis related to mobile Internet users broadly behavioral activities, including network heart activity, network social activity and network access activities; mainly related to the behavior of the users to access the network from the narrow sense law. By getting the behaviors of users form mobile Internet, telecom operators can effectively allocate network resources and to provide a targeted network services. Therefore, the establishment of a comprehensive mobile Internet user behavior analysis method and provide a valid user behavior analysis program has become an urgent problem of the telecom operators in the mobile Internet business development.This thesis is mainly for mobile Internet user behavior analysis and actionable analytics study, based on analysis of the key technologies of the mobile Internet user behavior analysis, combining data mining techniques, text classification techniques and sample reduction and balancing techniques proposed samples reduction and density equalization algorithm based on KNN, and the scheme of the mobile Internet user behavior analysis and experimental programs. This work is mainly reflected in:(1) According to the analysis of the demand for mobile Internet users’interested habits, a samples reduction and density equalization algorithm based on KNN was designed. To meet user behavior analysis based on the algorithm and achieve a balanced reduction for density of the sample space, the algorithm has improved both efficiency and accuracy in the process of analysis of user interest in mobile Internet habits.(2) According to analysis method for mobile Internet user behavior, the thesis designed a set of mobile Internet user behavior analysis methods and steps, provided the methodological basis for the mobile Internet user behavior analysis. The combination of mobile internet user behavior analysis methods, steps and proposed user behavior analysis scheme based on samples reduction and density equalization algorithm, and given a specific experiment. From the mobile Internet users access the network, the three dimensions of the access network traffic, time and user interest habits to achieve the analysis of user behaviors, the experimental results show that the proposed method, algorithms and solutions in the mobile internet user behavior analysis process is effective and feasible.更多还原