Commenting on local politics: An analysis of YouTube video comments for local government videos

This study compares the content of transcripts of videos uploaded by local governments with the comments on those videos, utilizing three transformer-model-based techniques: summarization of the discourse content of video transcripts, topic modeling of summarized transcripts, and sentiment analysis of transcripts and of comments. The analysis shows that some types of video content, for example those dealing with music or education, are more likely to attract positive comments than content related to policing or government meetings. In addition to their potential relevance for local government outreach, the study may represent a viable exploratory method for comparison of online video content and written comments in the context of computational social science analyses of user interaction and commenting behavior.