Exploring Public Discourse on Environmental Issues through Twitter: A Topic Modeling and Sentiment Analysis Approach

Marziye Sarvarizade

Bakalářská práce

Abstract:

Public discourse on environmental issues has gained momentum with the rise of social media, where platforms like Twitter and Reddit serve as spaces for discussion, debate, and information exchange. This study employs Natural Language Processing (NLP) techniques, including Latent Dirichlet Allocation (LDA) for topic modeling and sentiment analysis, to examine how environmental topics are framed and perceived in digital conversations. By analyzing discussions on Twitter and Reddit, the research identifies key themes, sentiment trends, and the role of public engagement in shaping environmental awareness and policy discourse. Findings reveal that Twitter conversations are highly reactive and crisis-driven, particularly in response to extreme weather events and policy decisions, with dominant emotions such as fear and disgust. In contrast, Reddit fosters more structured, in-depth discussions that balance criticism with constructive engagement, reflecting a broader emotional spectrum, including curiosity and optimism. These differences highlight the complementary nature of both platforms—Twitter captures real-time public reactions, while Reddit enables sustained discourse and knowledge-sharing. By leveraging these insights, this study offers implications for policymakers, researchers, and environmental advocates in improving communication strategies, addressing misinformation, and fostering informed public engagement on environmental issues.

Keywords

Natural Language Processing, Topic Modeling, Sentiment Analysis, Twitter, Reddit, Environmental Issues

Jazyk práce: angličtina

Datum vytvoření / odevzdání či podání práce: 27. 3. 2025

Obhajoba závěrečné práce

• Obhajoba proběhla 28. 5. 2025
• Vedoucí: doc. Zhongbing Chen
• Oponent: Doudou Ba, externi

Česká zemědělská univerzita v Praze

Fakulta životního prostředí

Bakalářský studijní program:
Environmental Data Science