The Israel-Hamas conflict that began in 2023 has resulted in extensive casualties and destruction, sparking worldwide discussions and garnering significant media attention. Users' and news groups' opinions regarding the conflict on social media underscore strong opinions and biases. Distilling the facts of the conflict from biased opinions is difficult, but we believe linguistic approaches like Natural Language Process- ing (NLP) can help. We posit that NLP techniques such as unigram word analysis and predictive text modeling can ob- jectively analyze user opinion and media bias. To support this claim, we created accurate predictive models to classify media and user biases with over 0.9 AUC in two datasets based on thousands of user and newsgroup tweets about the conflict. Significant unigram findings include limited discus- sion among pro-Israel about Palestinian casualties and pro- Palestinian groups making little reference to the actions of Hamas. This study not only provides a methodology applied to this conflict but also serves as a use-case for how NLP can quantify user and news bias, thus allowing people to better objectively evaluate issues with differing opinions.

PDF