Studying bias in Twitter (X) Community Notes
Name
Simon Fox Kuuse
Abstract
With the frequency of content in need of fact-checking constantly rising, new approaches in the form of crowdsourced fact-checking are being tested. In this thesis, we aim to identify potential bias in one of Twitter's approaches, called Community Notes. Two datasets, comprising Community Notes and bias ratings of sources, are collected. We utilise lexical features, sentiment analysis, temporal analysis and keyphrase extraction. Our study shows a correlation between the sentiment and reaction to real-world events and the bias of a Community Note, suggesting that a potential bias is present.
Graduation Thesis language
English
Graduation Thesis type
Bachelor - Computer Science
Supervisor(s)
Uku Kangur, Roshni Chakraborty
Defence year
2024