Clickbait Exposed: Sentiment Analysis as a Clickbait Detection Method

Abstract

The thesis explores the effectiveness of sentiment analysis as a method for detecting clickbait in digital news articles. It investigates the hypothesis that clickbait can be identified by comparing the emotional tone of headlines with their corresponding articles, known as comparative sentiment discrepancy analysis. This method was applied using two sentiment analysis models, TextBlob and VADER, on datasets from reputable and non-reputable news sources. The study found that while sentiment analysis has potential in identifying clickbait, the definition of clickbait itself is complex and nuanced, suggesting that a more holistic approach incorporating multiple methods would yield better results.