Browse Datasets

MasakhaNEWS

News Topic Classification for 16 African Languages

topic-classification language-modeling text-classification ~20,000 articles CC-BY-4.0
Access Dataset Read Paper

Description

High-quality, human-annotated news articles for topic classification. Excellent for modeling formal language and news domains, offering cleaner data than web crawls.

Quick Start

Python 
from datasets import load_dataset

dataset = load_dataset("masakhane/masakhanews", "hau", split="train")
print(dataset['text'][:3], dataset['label'][:3])

Project Ideas

Domain-Specific LM

Train a domain-specific language model for African news

Intermediate

Headline Generator

Generate news headlines using n-grams based on body text

Beginner

News Classifier

Train a classifier to categorize Swahili news articles

Beginner

Ethical Considerations

  • News data may contain political biases specific to region and publication time