Analysis And Application Of Network Video Service Characteristic Based On Big Data

With the rapid development of the Internet, the scale of online video users is continuing to expand and the traffic data in the Internet is increasing rapidly, the era of big data has come. Youku, as one of the most representative video sites in China, whose monthly active users exceed one hundred million, is generating massive data of user logs. In the era of data DT(Data Technology), how to collect, store and analyze the massive traffic data, mining user's behavior patterns and personal preferences, design market promotion plan, have important significance for both video sites and users.This thesis summarizes the background and research status of video websites first, and then discusses the data processing and analyzing platform, which contains four function modules. We analysis the characteristics of online video service from three aspects. Firstly, we mainly study the overall situation of top 10 video websites of China and find the differences among them. Mean while, we take Youku as a specific video site and explore the law of user behavior in it. Secondly,we propose the EPBP_ML model to predict the future popularity of online videos based on early popularity evolution pattern. Thirdly, we mining video user mobile pattern using the oscillation reduced data cleansed by the SOL algorithm presented in this paper.Based on real Internet data, this thesis studies the value of online video traffic data. Summarizing the law of user behavior and the characteristics of video traffic, predicting the future popularity of online videos based on early popularity evolution pattern and future popularity burst prediction, mining the mobile pattern of video users, have significant meanings in the era of big data.