Machine Learning-Based Sentiment Analysis of Climate Change Discussions on Twitter

Abstract: A more nuanced and complicated public opinion on climate change discussions on Twitter is what we want to shed light on. This study makes use of state-of-the-art ML and NLP tools, most notably Support Vector Machines (SVM), to do sentiment analysis on a worldwide scale using a large dataset acquired from a trustworthy external source. Focusing on both positive and negative emotions, this research aims to identify the many ways in which people express themselves emotionally while discussing climate change. Critically examined are the veracity and practicality of the chosen third- party dataset, which was acquired via Kaggle and made feasible by a Canadian Innovation Foundation JELF grant that was bestowed onto Chris Bausch of the University of Waterloo. With a remarkable F1 score of 0.70, the SVM-based sentiment analysis shows how well the selected approach captures the nuanced nature of climate change arguments on Twitter. Legislators, groups aiming to reach a worldwide audience, and climate change activists may all benefit from the research's communication tactics. The complex link between public opinion on climate change and online discourse is better understood in this research, which makes use of an external dataset and the sentiment analysis component of the Support Vector Machines method. In conclusion, this work adds substantial new information to the expanding field at the intersection of social media dynamics and environmental awareness by demonstrating the efficacy of support vector machines (SVMs) in identifying sentiment subtlety in massive datasets.

Keywords: Climate, sentiment analysis, and complexity are related terms.