Master's Thesis Analysing Climate Change Information Campaigns on Online Social Networks
- 1 minMade under the supervision of Erik Hemberg and Jérôme Baudry, carried at the ALFA group at the CSAIL, MIT
Master’s Thesis: Analyzing Information Campaigns on Climate Change Discussions in Online Social Networks
Overview
The master’s thesis delves into information campaigns. It analyzes the behavior of actors in these campaigns and assessing the information they disseminate.
Goal
The primary objective of this research is to provide an understanding of the dynamics of information flow in online social networks in the context of climate change discussions. The study aims to shed light on how information is spread, the activities of actors involved, and the credibility of content generated by Large Language Models (LLMs) in this domain.
Methodology
Transfer Entropy, an information theory-based metric, is used to identify and describe the behavior of actors involved in climate change discussions on OSNs. This method allows for the detection of information flow and simultaneous activities, providing insights into how information spreads in these networks.
In addition to this, the thesis explores the capabilities and limitations of Large Language Models (LLMs) in generating climate-related content. The analysis questions whether these models can produce articles with credibility comparable to those written by humans and highlights the presence of biases in the content generated by LLMs.
Interest
This research is of interest to practitioners engaged in media influence, social network analysis, and combating misinformation related to climate change. By presenting a novel methodological approach to investigate information campaigns on OSNs, the study provides critical insights into the actors and narratives at play in climate change discussions.
Applications
The insights derived from this thesis can be valuable in shaping strategies for information dissemination and combating misinformation related to climate change. Additionally, the understanding of the behavior of actors and the credibility of content generated by LLMs can assist in developing more effective communication and intervention strategies in the digital space. Furthermore, the study contributes to the broader field of social network analysis and information dissemination in online environments.
Links
- Link to the Thesis