基于大数据的网络视频业务特性的分析与应用

发布时间：2018-08-30 13:28

【摘要】：随着互联网的快速发展,网络视频用户规模不断扩大,互联网中的流量数据也在迅速膨胀,大数据时代随之到来。优酷作为国内最具代表性的视频网站之一,其业务渗透率已超过60%,月活跃用户数过亿,如此巨大的用户群体,必将产生海量的用户日志数据。在“数据DT时代”,如何采集、存储、分析这些海量数据,并从中挖掘出用户的行为模式及个人喜好,进而设计具有针对性的推广方案,对于视频网站和视频用户都有着显著的意义。论文首先对视频网站的背景和研究现状作了综述,进而介绍了由四个功能模块组成的网络视频业务特性分析平台。课题主要从三个方面对网络视频业务的特性进行了分析和研究。一是,分析移动互联网视频业务特性的总体情况,对比国内主流视频网站整体概况的差异,总结出国内热门的视频网站,并具体对优酷网站进行详细研究,总结用户行为规律。二是,基于视频播放量的增长模式,提出EPBP_ML模型对视频未来的播放量进行预测。三是,基于SOL算法去除噪声后的数据,提出Sequential Apriori模式挖掘算法来挖掘用户访问视频网站时的移动模式。基于真实的互联网流量数据,论文通过数据分析和数据挖掘来深入剖析网络视频数据背后蕴藏的价值,总结用户的行为规律和视频网站的流量特征、预测视频的播放量、挖掘用户的移动模式,在网络数据时代有着极其重要的意义。
[Abstract]:With the rapid development of the Internet, the scale of the network video users is expanding, and the traffic data in the Internet is expanding rapidly. Youku, as one of the most representative video websites in China, has more than 60 million users and more than 60 million active users every month. With such a huge user group, it will produce a huge amount of user log data. In the "data DT era", how to collect, store and analyze these massive data, and excavate the user's behavior patterns and personal preferences, and then design a targeted promotion scheme. For video sites and video users have a significant significance. Firstly, this paper summarizes the background and research status of video websites, and then introduces the analysis platform of network video services, which is composed of four functional modules. This paper mainly analyzes and studies the characteristics of network video service from three aspects. The first is to analyze the overall situation of mobile Internet video service, compare the differences of the domestic mainstream video websites, summarize the popular video websites in China, and make a detailed study on Youku station, and summarize the rules of user behavior. Secondly, based on the growth mode of video playback, the EPBP_ML model is proposed to predict the future video playback. Thirdly, based on the noisy data of SOL algorithm, a Sequential Apriori pattern mining algorithm is proposed to mine the mobile patterns of users visiting video websites. Based on the real Internet traffic data, this paper analyzes the value behind the network video data through data analysis and data mining, summarizes the behavior rules of users and the traffic characteristics of video websites, and predicts the amount of video playback. It is of great significance to excavate the mobile mode of users in the era of network data.
【学位授予单位】：北京邮电大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP311.13;TP393.092

【相似文献】