In March of this year, the Youtube channel JaidenAnimationsVideo uploaded a video titled ‘Being Not Straight’. The animated story, which has since received over 13 million views, is a coming out video that introduces the audience to aromanticism and asexuality. Common operationalisations include self-identification and the experience of little to no romantic and/or sexual attraction, desire, fantasies and/or interest. As both of these orientations are still relatively unknown, our project seeks to analyse the comment section under the video so as to investigate how the video was received. More precisely, we are interested in the audience's sentiment and the main topics discussed by the viewers.
To address our research aims, we use a combination of web scraping, sentiment analysis and topic modelling. We start our project by scraping the comments via the Youtube API. We then clean and explore the data. For this, we use a range of methods, namely subsetting, noise removal, negation bigrams, stopword removal, collocations, lemmatisation and the removal of unmeaningful frequency words. After that, we conduct a dictionary-based sentiment analysis to quantify how the majority of viewers felt about the video (i.e., positively or negatively). This process consists of two steps: First, we carry out a sentiment analysis with the vader package and a commonly used dictionary. Second, we account for any differences in the comments' popularity by weighting each comment according to their like number. Finally, to explore the key themes in the comments, we employ topic modelling.