Characterizing the correlation between video types and user quality of experience in the large-scale internet video service

The video service is one of the most important Internet services, and user quality of experience (QoE) represents the degree of recognition of the video service. Therefore, research on QoE is significant in improving the quality of the video service. Considering the influence of different factors on QoE, this paper studies the effect of video content types on QoE. In this article, we divide videos into seven types: entertainment, cartoon, sport, news, music, teleplay, specialty. Besides, we utilize user engagement to measure QoE. The study shows that the correlation which is mainly reflected in following aspects between video types and QoE is substantial. Firstly, video types can impact QoE directly. For example, the proportion of high engagement of cartoon is higher than other videos by 30%. Secondly, video types affect the distribution of the key performance indicators (KPIs). Lastly, video types influence the correlation between user engagement and KPI.