Investigating leadership discussion on social media: A NLP and machine learning perspective

The explosion of digital world has created a unique avenue for modern day social scientists and researchers that has allowed them to radically assess the relationships between people, as well as act as prosumers to these relationships as well. Compared to traditional media, social media is data and information rich, which has helped exploring human behavior at an unprecedented scale. A particular area that is being explored is leadership and leader usage of social media. This study tries to investigate a user's social media behavior surrounding leaders and leadership as a whole. It uses 54,326 tweets extracted using preferred keywords to perform analysis to evaluate the semantic orientation of these tweets and predict the causative drivers through generalized linear modeling for user's behavior. The findings suggest that Twitter users have a general positive sentiment towards leadership, and there are four distinct topics that explain this relationship. This was further validated through predictive analytics, and it showed that there a statistically significant impact of sentiment polarity on the user's behavior. This study proposes a scheme for evaluating the causative factors of Twitter user's behavior which provides us with insights into the use of social media for leaders and leadership.