Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

What distinguishes conspiracy from critical narratives? A computational analysis of oppositional discourse

Damir Korenčić, Berta Chulvi, Xavier Bonet Casals, Alejandro Toselli, Mariona Taulé, Paolo Rosso

TL;DR

This work tackles the problem of differentiating conspiracy theories from critical opposition in text by introducing a topic-agnostic annotation scheme that integrates inter-group conflict (IGC) elements. It presents the multilingual XAI-DisInfodemics corpus (English and Spanish Telegram messages about COVID-19) with binary conspiracy-vs-critical labels and span-level narrative annotations, enabling both text-level and span-level NLP tasks. Transformer-based classifiers achieve high accuracy on distinguishing conspiracy from critical content, while span-level detection of narrative elements is more challenging; analyses show conspiratorial discourse exhibits higher anger, violence, and IGC activity. The findings offer practical implications for fair content moderation and interdisciplinary research, and point to future work in expanding topic coverage, platforms, multilingual models, and learning-from-disagreement approaches.

Abstract

The current prevalence of conspiracy theories on the internet is a significant issue, tackled by many computational approaches. However, these approaches fail to recognize the relevance of distinguishing between texts which contain a conspiracy theory and texts which are simply critical and oppose mainstream narratives. Furthermore, little attention is usually paid to the role of inter-group conflict in oppositional narratives. We contribute by proposing a novel topic-agnostic annotation scheme that differentiates between conspiracies and critical texts, and that defines span-level categories of inter-group conflict. We also contribute with the multilingual XAI-DisInfodemics corpus (English and Spanish), which contains a high-quality annotation of Telegram messages related to COVID-19 (5,000 messages per language). We also demonstrate the feasibility of an NLP-based automatization by performing a range of experiments that yield strong baseline solutions. Finally, we perform an analysis which demonstrates that the promotion of intergroup conflict and the presence of violence and anger are key aspects to distinguish between the two types of oppositional narratives, i.e., conspiracy vs. critical.

What distinguishes conspiracy from critical narratives? A computational analysis of oppositional discourse

TL;DR

This work tackles the problem of differentiating conspiracy theories from critical opposition in text by introducing a topic-agnostic annotation scheme that integrates inter-group conflict (IGC) elements. It presents the multilingual XAI-DisInfodemics corpus (English and Spanish Telegram messages about COVID-19) with binary conspiracy-vs-critical labels and span-level narrative annotations, enabling both text-level and span-level NLP tasks. Transformer-based classifiers achieve high accuracy on distinguishing conspiracy from critical content, while span-level detection of narrative elements is more challenging; analyses show conspiratorial discourse exhibits higher anger, violence, and IGC activity. The findings offer practical implications for fair content moderation and interdisciplinary research, and point to future work in expanding topic coverage, platforms, multilingual models, and learning-from-disagreement approaches.

Abstract

The current prevalence of conspiracy theories on the internet is a significant issue, tackled by many computational approaches. However, these approaches fail to recognize the relevance of distinguishing between texts which contain a conspiracy theory and texts which are simply critical and oppose mainstream narratives. Furthermore, little attention is usually paid to the role of inter-group conflict in oppositional narratives. We contribute by proposing a novel topic-agnostic annotation scheme that differentiates between conspiracies and critical texts, and that defines span-level categories of inter-group conflict. We also contribute with the multilingual XAI-DisInfodemics corpus (English and Spanish), which contains a high-quality annotation of Telegram messages related to COVID-19 (5,000 messages per language). We also demonstrate the feasibility of an NLP-based automatization by performing a range of experiments that yield strong baseline solutions. Finally, we perform an analysis which demonstrates that the promotion of intergroup conflict and the presence of violence and anger are key aspects to distinguish between the two types of oppositional narratives, i.e., conspiracy vs. critical.
Paper Structure (15 sections, 5 figures, 7 tables)

This paper contains 15 sections, 5 figures, 7 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. Dataset: XAI-DisInfodemic Corpus
  4. Corpus Construction
  5. Conspiracy vs. Critical Annotation
  6. Oppositional Narrative Annotation
  7. Anonymization
  8. Predictive Machine Learning
  9. Classifying Critical vs. Conspiracy
  10. Detection of Narrative Elements
  11. Detection of Narrative Elements
  12. Error Analysis
  13. Applying the Annotation Scheme
  14. Analysis and Results
  15. Conclusions and Future Work

Figures (5)

  • Figure 1: The process of corpus construction.
  • Figure 2: A Conspiracy message annotated with elements of oppositional narrative: Agents (A), Facilitators (F), Campaigners (C), Victims (V), Objectives (O), Negative Effects (E).
  • Figure 3: A Critical message annotated with elements of oppositional narrative: Agents (A), Facilitators (F), Campaigners (C), Victims (V), Objectives (O), Negative Effects (E).
  • Figure 4: Relation of the conspiracy and critical label with the presence of anger and violence words in texts.
  • Figure 5: Relation of the IGC variable with the percentage of anger and violence words in messages.