Fattori determinanti le opinioni sul cambiamento climatico su Reddit

In recent years, climate change has become one of the most discussed topics in public discourse. For this reason, it is essential to study the dynamics of people's opinions on this issue. The purpose of this thesis is to create a probabilistic model using Natural Language Processing (NLP) techniques to study the opinions of social media users on climate change. Our focus is on the discussion website Reddit: the richness of its data allows to connect the opinion or the emergence of new interest about climate change with the underlying socio-demographic characteristics of users. To achieve this goal, a dataset containing millions of comments on this topic posted on the social network Reddit was created. After analyzing it, the messages were classified automatically with a Transformer deep learning model based on the opinions expressed on the issue of global warming. Subsequently, the characteristics that distinguish users in favor of global warming mitigation policies from those who are not interested in the topic were studied. It was found that the former are generally left-leaning voters, not too young or poor compared to Reddit users, and they are interested in scientific, political, and current affairs topics.