Exploring Online Anorexia (Pro-Ana) Communities : A Comprehensive Text Mining Exploration of Pro-Ana Discourses on Twitter

Pro-ana online communities on social media continue to celebrate anorexia. Previous research into pro-ana communities primarily relied on manual content analysis, limiting the quantity of analysed data. This study used automated text mining to identify discussed topics and sentiments and examined engagement metrics (likes, retweets, replies, quote count). The identified main themes were dieting, social engagement, and weight management, which aligned with prior research on anorectic preoccupations. Tweets exhibited a mainly neutral tone, with much less frequent positive and negative sentiments. Very weak but significant positive correlations were found between sentiment and likes, retweets, and replies, and a negative correlation was found with quote count. Notably, only a limited number of tweets actually showed any engagement according to the engagement metrics. This study confirms results from past manual content analysis of pro-ana communities,indicating a strong focus on content revolving around slim appearance, weight, and food management. It also expands knowledge of social media discourses supporting typically anorectic behaviour. The sentiment analysis suggests rather informative content-wise and less emotionally charged tweets. Emotionally charged tweets that, if present, evoke the most engagement. Due to the weak relationships with sentiment and engagement metrics, exploring other factors that promote interaction would be essential.