NLP-Based Sentiment Analysis of Social Media Data on Public Perception of Environmental Policies

Abstract

The future of any environmental policy all depends on the view of the citizens since the support of its citizens directly acquires the outcome of such a policy; whether to be successful, or not in terms of compliance, political will and long term sustainability. The social media expansion has been more visible and disorganized, polarized and fragmented because of public opinion. The proposed research is based on the concepts of Natural Language Processing (NLP) to conduct a bulk of sentiment analysis of social media information regarding the environmental-related policy debate of carbon taxes, plastic bans and the incorporation of renewable energy. It was chosen to follow the hybrid research methodology approach that would involve both the sentiment classification via lexicon and the machine learning algorithms, including Support Vector Machines (SVM) and deep learning-based Bidirectional Encoder Representations of Transformers (BERT). The social media variables were collected on Twitter (X) and Facebook in 2022-2024, and covered over 1.2 million posts, comments, and hashtags on policy debates. The results indicate that the correlation between the popular view and the policy acceptance is high with the policies on renewable energy having predominantly positive sentiment ( +62) and the carbon tax proposals having negative sentiment ( -54) referring to the affordability aspect. The temporal trend analysis indicated that there were occurrences of negative sentiment when announcements of political nature and environmental protests were high. The findings indicate the usefulness of NLP-informed sentiment analysis in providing a policy holder with actionable information, policy-based communication, improved stakeholder engagement, and responsive policymaking within the environmental context. The article demonstrates that computational social science can be used alongside existing methods of policy evaluation to offer scalable and real-time monitoring of the social opinion regarding environmental governance.