Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Tackling Social Bias against the Poor: A Dataset and Taxonomy on Aporophobia

Georgina Curto, Svetlana Kiritchenko, Muhammad Hammad Fahim Siddiqui, Isar Nejadgholi, Kathleen C. Fraser

TL;DR

This paper tackles aporophobia, the bias against the poor, by operationalizing it for NLP and social media analysis. It introduces DRAX, the first manually annotated dataset of 1,816 English tweets from multiple regions, and a taxonomy of aporophobic actions categorized into Direct and Reporting signals. It evaluates automatic detection using RoBERTa and GPT-4o, finding moderate performance (best F1 around 0.64) and showing that general toxicity detectors underperform for this task. The work enables scalable tracking of aporophobia and motivates region-aware data collection and specialized models to inform policy and interventions.

Abstract

Eradicating poverty is the first goal in the United Nations Sustainable Development Goals. However, aporophobia -- the societal bias against people living in poverty -- constitutes a major obstacle to designing, approving and implementing poverty-mitigation policies. This work presents an initial step towards operationalizing the concept of aporophobia to identify and track harmful beliefs and discriminative actions against poor people on social media. In close collaboration with non-profits and governmental organizations, we conduct data collection and exploration. Then we manually annotate a corpus of English tweets from five world regions for the presence of (1) direct expressions of aporophobia, and (2) statements referring to or criticizing aporophobic views or actions of others, to comprehensively characterize the social media discourse related to bias and discrimination against the poor. Based on the annotated data, we devise a taxonomy of categories of aporophobic attitudes and actions expressed through speech on social media. Finally, we train several classifiers and identify the main challenges for automatic detection of aporophobia in social networks. This work paves the way towards identifying, tracking, and mitigating aporophobic views on social media at scale.

Tackling Social Bias against the Poor: A Dataset and Taxonomy on Aporophobia

TL;DR

This paper tackles aporophobia, the bias against the poor, by operationalizing it for NLP and social media analysis. It introduces DRAX, the first manually annotated dataset of 1,816 English tweets from multiple regions, and a taxonomy of aporophobic actions categorized into Direct and Reporting signals. It evaluates automatic detection using RoBERTa and GPT-4o, finding moderate performance (best F1 around 0.64) and showing that general toxicity detectors underperform for this task. The work enables scalable tracking of aporophobia and motivates region-aware data collection and specialized models to inform policy and interventions.

Abstract

Eradicating poverty is the first goal in the United Nations Sustainable Development Goals. However, aporophobia -- the societal bias against people living in poverty -- constitutes a major obstacle to designing, approving and implementing poverty-mitigation policies. This work presents an initial step towards operationalizing the concept of aporophobia to identify and track harmful beliefs and discriminative actions against poor people on social media. In close collaboration with non-profits and governmental organizations, we conduct data collection and exploration. Then we manually annotate a corpus of English tweets from five world regions for the presence of (1) direct expressions of aporophobia, and (2) statements referring to or criticizing aporophobic views or actions of others, to comprehensively characterize the social media discourse related to bias and discrimination against the poor. Based on the annotated data, we devise a taxonomy of categories of aporophobic attitudes and actions expressed through speech on social media. Finally, we train several classifiers and identify the main challenges for automatic detection of aporophobia in social networks. This work paves the way towards identifying, tracking, and mitigating aporophobic views on social media at scale.

Paper Structure

This paper contains 20 sections, 4 figures, 7 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. DRAX Dataset Collection and Annotation
  4. Tweet Collection and Data Sampling
  5. Data Annotation
  6. Characteristics of the DRAX Dataset
  7. Taxonomy of Aporophobic Actions Expressed Through Speech
  8. Automatic Aporophobia Detection
  9. Discussion and Future Work
  10. Details on the Dataset Creation
  11. Tweet Collection and Pre-processing
  12. Topic Modeling with BERTopic
  13. Human Annotators
  14. Data Distribution
  15. Additional Details and Experiments with NLP Models
  16. ...and 5 more sections

Figures (4)

  • Figure 1: The process diagram for the novel methodology to collect and annotate data.
  • Figure 2: Class distribution in DRAX per geographical region.
  • Figure 3: Taxonomy of categories for three levels of classification of aporophobia: Type of speech, Degree of action, Categories of aporophobia expressed through speech.
  • Figure A.1: Class distribution per topic in the DRAX dataset.