Multilingual African News Articles for Topic Classification

MasakhaNEWS provides news articles for topic classification across 16 widely spoken African languages. The dataset was created by the Masakhane community and includes train, validation, and test splits for all languages. It was last updated in December 2025.

Use Cases

• Train a text classifier to predict news topics (e.g., business, sport, politics) from article text in languages like Amharic or Swahili.
• Benchmark multilingual model performance on the topic classification task across the 16 provided language splits.
• Fine-tune a language model for cross-lingual transfer learning using the parallel topic labels across different languages.
• Analyze linguistic or topical trends in African news media by comparing classifications across the language-specific datasets.

Strengths

• Largest publicly available dataset for news topic classification in 16 African languages.
• Includes standardized train, validation, and test splits for all 16 languages.

Limitations

• Specific row counts, column details, and sample sizes are not provided in the available description.
• The geographic scope and temporal coverage of the collected news articles are unknown.

Provenance

Source

Masakhane community.

Collection Method

Expert-generated annotations for news topic classification.

Freshness

Last updated in December 2025.

Geography

Covers 16 widely spoken African languages.