Home // HUSO 2023, The Ninth International Conference on Human and Social Analytics // View article

Examining Content and Emotion Bias in YouTube’s Recommendation Algorithm

Authors:
Obianuju Okeke
Mert Can Cakmak
Billy Spann
Nitin Agarwal

Keywords: Recommender Systems; Recommendation Bias; YouTube; Topic Modeling; Emotion Modeling.

Abstract:
Detection, characterization, and mitigation of bias in modern systems of automated and autonomous decisions is a growing interdisciplinary field. This study aims to explore YouTube’s video recommendation bias to determine if an inherent bias has an unintended impact of occluding vulnerable communities and minority groups. Our findings suggest that the algorithm recommends videos evoking more positive emotions and higher user engagement. We also discovered that content related to our seed videos was filtered out in a systematic but gradual pendulum-like motion. This analysis of potential emergent biases will be applicable in the fairness of recommender systems, patterns of content consumption, information diffusion, echo-chamber formation, and other significant problems.

Pages: 15 to 20

Copyright: Copyright (c) IARIA, 2023

Publication date: March 13, 2023

Published in: conference

ISSN: 2519-8351

ISBN: 978-1-68558-066-7

Location: Barcelona, Spain

Dates: from March 13, 2023 to March 17, 2023