Investigating Bias in YouTube Recommendations: Emotion, Morality, and Network Dynamics in China-Uyghur Content

This research demonstrates how recommendation algorithms can shape public discourse and attitudes on morally charged topics. Our study provides an in-depth analysis of the moral, emotional, and network dimensions of Youtube’s recommended videos related to the China-Uyghur issue. We investigated the spread of moral themes and whether the algorithm favors videos with certain emotional feelings. Additionally, we conducted a detailed network analysis to spot the most influential videos and see how the themes change as the recommendations change. We found that as the algorithm recommends more videos, the emotional diversity of the recommendations tends to drift towards positive emotions and away from negative ones. Likewise, there is a decreasing focus on moral dilemmas as one moves through the recommended content. In simple terms, our study shows how YouTube’s recommendations may influence viewers’ feelings and beliefs. Our network analysis reveals which videos are driving the shift in morality and emotion and how the main discussion points change as more videos are suggested. Through this, we hope to better understand the inherent biases in recommendation engines, especially when they are dealing with emotionally charged and morally complex topics.