Online videos have become the most popular method to obtain information for the public in recent years, such as Tiktok, YouTube, and regional sites Bilibili, Douyin, etc. Compared with its growing influence, the analysis of user behavior of video sites is still less investigated. Here we fetch the video data on bilibili.com, and analyze the video plays, comments and other behaviors on the website. We found that the prediction model based on the Hawkes process can accurately predict the video view counts, which suggests that on the Bilibili website the self-incentive mechanism of information cascade diffusion plays a decisive role in online views. Meanwhile, we also found that the views increment of the videos during the same period of time conforms to the general power-law distribution.
In this repository, we uploaded the data we analyzed in our paper.