Using Artificial Intelligence to Monitor the Evolution of Opinion Leaders’ Sentiments: Case Study on Global Warming

Abstract

Emotional and sentiment analysis of social media content is essential for smart city analytics. In the past few years, researchers have relied on online content's sentiment analysis to capture public opinion about current events. Despite their merits, the existing solutions take a retrospective coarse-grained approach that analyses millions of social media posts to study public opinion about past events (e.g., presidential elections). Such models give late insights, which makes it difficult to intervene or adjust strategies based on the evolution of public opinion over time. Moreover, such approaches lack efficiency and scalability since they require the analysis of millions of posts to obtain accurate results. In this work, we address those limitations by proposing a framework for the real-time monitoring of the evolution of public opinion over time. To ensure efficiency and scalability, we focus on the analysis of high impact social media content generated by opinion leaders and their followers. To build our framework, we leveraged our opinion leaders' identification algorithm, along with text mining and text classification techniques, to capture and analyze the evolution of the sentiments and emotions of 34 opinion leaders concerning the topic of global warming. The results obtained are very promising and open the door to advanced social media analytics to monitor public opinion in real-time.