Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Auditing health-related recommendations in social media: A Case Study of Abortion on YouTube

Mohammed Lahsaini, Mohamed Lechiakh, Alexandre Maurer

TL;DR

This paper tackles the problem of health misinformation and bias in YouTube recommendations, focusing on abortion content. It introduces a sock-puppet auditing framework to treat YouTube as a black-box recommender system and to study how recommendations vary with simulated user profiles. By building recommendation graphs and applying multiple centrality metrics, the authors identify influential videos and assess bias and misinformation spread, finding a dominance of pro-abortion and debunk content among the top recommendations with limited misinformation exposure overall. The work contributes a scalable, practical auditing approach applicable to other platforms and topics, with implications for platform design, policy, and safeguarding health information online.

Abstract

Recommendation algorithms (RS) used by social media, like YouTube, significantly shape our information consumption across various domains, especially in healthcare. Hence, algorithmic auditing becomes crucial to uncover their potential bias and misinformation, particularly in the context of controversial topics like abortion. We introduce a simple yet effective sock puppet auditing approach to investigate how YouTube recommends abortion-related videos to individuals with different backgrounds. Our framework allows for efficient auditing of RS, regardless of the complexity of the underlying algorithms

Auditing health-related recommendations in social media: A Case Study of Abortion on YouTube

TL;DR

This paper tackles the problem of health misinformation and bias in YouTube recommendations, focusing on abortion content. It introduces a sock-puppet auditing framework to treat YouTube as a black-box recommender system and to study how recommendations vary with simulated user profiles. By building recommendation graphs and applying multiple centrality metrics, the authors identify influential videos and assess bias and misinformation spread, finding a dominance of pro-abortion and debunk content among the top recommendations with limited misinformation exposure overall. The work contributes a scalable, practical auditing approach applicable to other platforms and topics, with implications for platform design, policy, and safeguarding health information online.

Abstract

Recommendation algorithms (RS) used by social media, like YouTube, significantly shape our information consumption across various domains, especially in healthcare. Hence, algorithmic auditing becomes crucial to uncover their potential bias and misinformation, particularly in the context of controversial topics like abortion. We introduce a simple yet effective sock puppet auditing approach to investigate how YouTube recommends abortion-related videos to individuals with different backgrounds. Our framework allows for efficient auditing of RS, regardless of the complexity of the underlying algorithms
Paper Structure (26 sections, 3 figures, 3 tables)

This paper contains 26 sections, 3 figures, 3 tables.

Table of Contents

  1. Introduction
  2. Background
  3. Motivation
  4. Objective
  5. Literature review
  6. Misinformation and bias in recommendation systems
  7. In the context of health-related content
  8. Auditing YouTube recommendations
  9. Material and methods
  10. Data collection
  11. Data collection process
  12. Simulated profiles
  13. Search queries
  14. Video selection
  15. Data annotation and analysis
  16. ...and 11 more sections

Figures (3)

  • Figure 1: Our sock puppet auditing system to measure bias and the spread of misinformation of abortion-related content on YouTube.
  • Figure 2: Recommendations graphs for the six profiles. The node's size and color are correlated with the length and number of views of the video, respectively.
  • Figure 3: Classification of the most influential videos in each graph.