Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Narratives at Conflict: Computational Analysis of News Framing in Multilingual Disinformation Campaigns

Antonina Sinelnik, Dirk Hovy

TL;DR

It is found that disinformation campaigns consistently and intentionally favor specific framing, depending on the target language of the audience, and how Russian-language articles consistently highlight selected frames depending on the region of the media coverage is discovered.

Abstract

Any report frames issues to favor a particular interpretation by highlighting or excluding certain aspects of a story. Despite the widespread use of framing in disinformation, framing properties and detection methods remain underexplored outside the English-speaking world. We explore how multilingual framing of the same issue differs systematically. We use eight years of Russia-backed disinformation campaigns, spanning 8k news articles in 4 languages targeting 15 countries. We find that disinformation campaigns consistently and intentionally favor specific framing, depending on the target language of the audience. We further discover how Russian-language articles consistently highlight selected frames depending on the region of the media coverage. We find that the two most prominent models for automatic frame analysis underperform and show high disagreement, highlighting the need for further research.

Narratives at Conflict: Computational Analysis of News Framing in Multilingual Disinformation Campaigns

TL;DR

It is found that disinformation campaigns consistently and intentionally favor specific framing, depending on the target language of the audience, and how Russian-language articles consistently highlight selected frames depending on the region of the media coverage is discovered.

Abstract

Any report frames issues to favor a particular interpretation by highlighting or excluding certain aspects of a story. Despite the widespread use of framing in disinformation, framing properties and detection methods remain underexplored outside the English-speaking world. We explore how multilingual framing of the same issue differs systematically. We use eight years of Russia-backed disinformation campaigns, spanning 8k news articles in 4 languages targeting 15 countries. We find that disinformation campaigns consistently and intentionally favor specific framing, depending on the target language of the audience. We further discover how Russian-language articles consistently highlight selected frames depending on the region of the media coverage. We find that the two most prominent models for automatic frame analysis underperform and show high disagreement, highlighting the need for further research.
Paper Structure (23 sections, 1 equation, 5 figures, 7 tables)

This paper contains 23 sections, 1 equation, 5 figures, 7 tables.

Table of Contents

  1. Introduction
  2. Contributions:
  3. Data
  4. Generating Article Pairs
  5. Method and Modeling
  6. Method Comparison Overview
  7. Lexicon-based Frame Identification
  8. Methodology
  9. Evaluating the Lexicon
  10. Transformer-based Frame Identification
  11. Methodology
  12. Evaluating the Model
  13. Evaluating and Comparing Models
  14. Introduction
  15. Analysis of Competence and Agreement on Dominant Frames
  16. ...and 8 more sections

Figures (5)

  • Figure 1: % of annotations where two methods reach agreement about frame's presence, by language
  • Figure 2: PMI score for four languages, normalized to [-1;1]
  • Figure 3: PMI score for four regions, normalized to [-1;1]
  • Figure 4: Keywords Cosine Similarity for a Pair of Ground Truth Articles
  • Figure 5: Normalized Confusion Matrix; the codes represent the frames, see code-frame correspondence in Table \ref{['tab:xlmr_train_test_counts']} or Table \ref{['tab: f1_per_frame']}