12,000 news articles from iLur.am categorized into seven classes: sport, politics, weather, economy, accidents, art, and society. The corpus is divided into a training set of 2,242,000 tokens and a test set of 425,000 tokens.
Use Cases
- Train a multi-class text classifier using the 7 provided category labels
- Perform linguistic analysis or language modeling on the 2.6 million tokens of Armenian news text
- Benchmark news categorization algorithms using the predefined train and test splits
Strengths
- 12,000 news articles sourced from the iLur.am news portal
- 7 categorical labels including sport, politics, weather, economy, accidents, art, and society
- 2,242,000 tokens in the training partition and 425,000 tokens in the test partition