Search

Browse

• Communities
• Authors
• Year
• Titles
• Subjects

My Space

• Login

Detecting and analyzing false claims for businesses and consumers using large language models

Kyvernitakis-Synanis Antonios

Full record

---

Summary

Misinformation, primarily manifested through fake news, poses an increasing threat to the smooth operation of modern societies. Concerning the business and consumer ecosystem, the spread of fake news damages the reputation of businesses, undermines the trust of consumers, and affects their decisions. This thesis focuses on analyzing false claims targeting businesses and/or consumers, aiming to understand their characteristics, dissemination mechanisms, and impacts, as well as evaluating the effectiveness of modern artificial intelligence technologies for automating their detection and analysis. Initially, a comprehensive literature review on online misinformation and its effects on businesses and consumers at international level is conducted. The research then focuses on the Greek internet, using data from the platform ellinikahoaxes.gr to find and analyze false claims concerning businesses and/or consumers. By systematically extracting data from ellinikahoaxes.gr, the thesis employs a Large Language Model, specifically ChatGPT 4o, to investigate the effectiveness of this tool in two natural language processing tasks (over texts in Greek): a) detecting claims related to businesses and/or consumers, and b) assessing the check-worthiness of a piece of text. Through this evaluation, the thesis concludes with findings and recommendations on the use of advanced machine learning and natural language processing technologies in combating misinformation to protect the interests of both businesses and consumers.

Available Files

Services