Discovering fans and anti-fans among social media users based on their emotional reactions and comments

Abstract

While audience profiling has been a critical concern in social media marketing, little research has used a systematic methodology to identify fans and anti-fans in social media communities. This study aimed to develop a fan and anti-fan detection model by analysing social media users’ mood responses and comments on fan page posts. The sentiment analysis of comments was conducted using a bidirectional long short-term memory (LSTM) model. A total of 83 posts, 849,820 emoticons and 216,688 comments were collected from two different fan pages over 14 days. Results showed that the proposed model, combining emotional reaction analysis and sentiment analysis of their comments, exhibited better fan and anti-fan identification capability than the single-dimensional behaviour model. It exhibited 96% accuracy, 100% precision and 93% recall in terms of community management. This study provides a novel, accurate and efficient way to identify fans and anti-fans that can help form more targeted marketing strategies for social media managers in a cost-effective manner.