A measurement study of online internet video views through a large-scale, search engine

As the spread of broadband Internet access, the online Internet video becomes popular in our daily life. The search engine is an important tool for users to access online video, resulting in considerable traffic on the video Websites. In this work, we statistically analyze the video access log provided by a large scale video search engine, and validate that the aggregated views of online video from the search engine is diurnally regular. Furthermore, we find the probability of abrupt surge or disappearance of views is significantly high, having a percentage of 20% all videos. The results also show that 30% of them are spike that indicates that an accessing surge of views is followed by an immediate and abrupt falling. Such a result renders the big challenge how the Website providers can satisfy the quality of service in providing their videos to users during the surging period. The online video access also indicates the social behaviors of the corresponding social group. These results are revealed by studying the querying string at the search engine. These results can be useful for the Website developer to optimize the video system by allocating appropriate system resources according to the accessing habits of the users.