大数据视角下的互联网视频用户行为

Internet Video User's Behavior from Perspective of Big Data

短短10余年间,互联网视频便以其强娱乐性、高个性化、高质量的特点征服了广大用户。规模较大的互联网视频服务网络往往拥百万级的用户数,其每日可采集到数百万条用户行为数据。通过挖掘用户行为数据,发现隐蔽其中的用户行为规律,是提高互联网视频运营企业节目生产、投放能力和改善运营服务能力的有效手段。本文中,我们以最小二乘法、相关性分析法和信息增益法为数学工具,对国内某大型互联网视频运营商采集的用户行为数据进行了深度挖掘。通过研究,我们发现了一系列的用户行为规律,并对其进行了归纳,还对影响用户行为和用户参与度的诸多因素进行了分析。

During the ten years, Internet video has conquered users with strong entertainment, high individualization and high-quality. A mainstream Internet video service provider usually holds millions of users, and collects millions pieces of user＇s behavior data every day. To find out user＇s behavior patterns by mining the user＇s behavior big data is an effective means for Internet video provider to improve capacities of production, delivery and service. In this paper, we utilize mathematical tools, such as least square method, correlation analysis and information gain, for the purpose of mining the user＇s behavior data collected by one of the large native Internet video providers. We have found out a series of user＇s behavior patterns and summed them up. Also we analyze the factors that affect the user＇s behavior and engagement.