Mapping Affective Polarization in YouTube Shorts: A Data-Driven Analysis of Political Communication During the 2023–2024 Israel–Hamas War

Abstract

This study examines the application of computational models, particularly Aspect-Based Sentiment Analysis (ABSA), to the analysis of political communication in online spaces. The paper provides an in-depth analysis of affective polarization in user-generated content on YouTube Shorts related to the 2023–2024 Israel–Hamas war. By analyzing a dataset comprising more than 3.4 million comments and replies, the study utilizes a fine-tuned DeBERTa-v3-large-absa-v1.1 model to trace subtle, nuance-driven sentiment pathways that reveal ideological inclinations toward geopolitical actors. The findings indicate a persistent negative sentiment toward Israel and Zionists, alongside a consistent positive sentiment toward Palestine and Palestinians. Overall, the study illustrates how affective polarization emerges as a reaction to state-funded media discourse. Beyond its methodological contributions, the paper addresses domain adaptation, demonstrating that language imbued with evaluative and emotional undertones can be analyzed and quantitatively assessed to characterize political discourse within Web 2.0 ecosystems. Furthermore, it connects these dynamics to broader discussions surrounding polarization in digital mainstream environments.