Nepali News Articles with Category Labels

Over 2.76 million Nepali news articles scraped from Baahrakhari and other sources, with cleaned category labels. The dataset was created by spandyie and was last updated in February 2026. It is provided in Parquet format compressed with Snappy.

Use Cases

• Train a text classifier to predict the `category` label from the `heading` and `text` columns.
• Analyze trends in news coverage over time using the `date` field and article `text`.
• Build a Nepali language model by pre-training on the large corpus of `text` from news articles.
• Perform sentiment or topic analysis on the `heading` and `text` of Nepali news content.

Strengths

• Contains approximately 2.76 million articles, providing a substantial corpus for NLP tasks.
• Includes cleaned `category` labels for supervised learning applications.

Limitations

• The `date` field uses the Nepali calendar, which may require conversion for standard temporal analysis.
• The dataset's geographic and source bias is unknown, potentially limiting representativeness of all Nepali news.

Provenance

Source

Scraped from Baahrakhari and other Nepali news sources.

Collection Method

Web scraping.

Freshness

Last updated on 2026-02-23.

Geography

Nepal