KinyaProp: Fine-grained Propaganda Analysis in Kinyarwanda

License:

This repository contains KinyaProp, a fine-grained propaganda dataset annotated on Kinyarwanda news articles.

---

File Structure

KinyaProp/
├── README.md
├── dataset/
│   ├── annotations_raw.json
│   └── gold_majority_vote.json

---

File Description

1) annotations_raw.json

This file contains raw annotations independently produced by three annotators.

Each entry has the following structure:

[
  {
    "article_num": "article ID",
    "article_text": "article title\n\n article content",
    "category": "article category ID",
    "kin_label": "category name in Kinyarwanda",
    "annotators": {
      "annotator_id": [
        {
          "text": "propagandistic span text",
          "label": "propaganda technique label",
          "start": span_start,
          "end": span_end
        }
      ]
    }
  }
]

Category mapping:

• 1 → politics
• 2 → sport
• 3 → economy
• 4 → health
• 5 → entertainment
• 6 → history
• 7 → technology
• 8 → tourism
• 9 → culture
• 10 → fashion
• 11 → religion
• 12 → environment
• 13 → education
• 14 → relationship

---

2) gold_majority_vote.json

This file contains final (gold) annotations produced using a majority agreement strategy.

Majority agreement: A span is assigned a label when at least two annotators agree.

Each entry has the following structure:

[
  {
    "article_num": "article ID",
    "article_text": "article title\n\narticle content",
    "category": "article category ID",
    "kin_label": "category name in Kinyarwanda",
    "gold_spans": [
      {
        "text": "propagandistic span text",
        "label": "propaganda technique label",
        "start": span_start,
        "end": span_end
      }
    ]
  }
]

---

Citation

If you use this dataset, please cite:

Manzi Fabrice Niyigaba, Ivory Yang, and Soroush Vosoughi. 2026. KinyaProp: Fine-Grained Propaganda Annotation in Kinyarwanda. In Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 34230-34243, San Diego, California, United States. Association for Computational Linguistics.