Public Opinion Analysis Strategy of Short Video Content Review in Big Data Environment

Abstract

With the rapid development of mobile Internet and 4G, short video apps came into being. Internet companies have launched such explosive short video platforms as weishi, TikTok and kuaishou in response to the fragmented reading habits of the public. Internet users can more easily browse public opinion news, express opinions and emotions. In view of a variety of public opinion events and a large amount of comment information from Internet users, this paper proposes a big data public opinion analysis strategy which integrates short video content comments. Firstly, an improved kernel k-means algorithm based on local density and single-pass is proposed for topic discovery, which solves the problems of uncertainty of initial center point and high time complexity of K-means algorithm. Then, according to the characteristics of online public opinion, a model is proposed to quantitatively express the emotional value of public opinion comments. In the spark platform, the description titles and high praise comments of the hot videos of the short video platform in the past three months are used as the data set for simulation experiments. The results show that the improved algorithm improves the clustering effect, and solves the low efficiency problem caused by the high time complexity and large amount of data. The affective index of hot events given by the affective value measurement model accords with the public opinion guidance of recent public opinion events.