Sinhala News Texts with Five Category Labels

A collection of Sinhala language news texts belonging to 5 categories: political, business, technology, sports, and entertainment. The data was originally released by Nisansa de Silva in 2015 and has been processed to remove single-word texts and English-only sentences.

Use Cases

• Train a text classifier to predict the 5 news categories (political, business, technology, sports, entertainment) from Sinhala news sentences.
• Analyze linguistic patterns and vocabulary differences across the 5 distinct Sinhala news categories.
• Benchmark multilingual NLP models on a resource-poor language using this processed Sinhala text corpus.

Strengths

• Data is cleaned of single-word texts and English-only sentences.
• Provides a labeled corpus for a resource-poor language (Sinhala).
• Contains texts from 5 distinct and common news categories.

Limitations

• The original dataset is from 2015, so the language and topics may be temporally stale.
• The exact size (row count) and specific features (columns) of the processed dataset are unknown.
• Potential class imbalance is not described.

Provenance

Source

Originally released by Nisansa de Silva (2015), processed and hosted by NLPC-UOM.

Collection Method

Crowdsourced, as indicated by platform tags.

Time Range

Original data from 2015.

Freshness

Last updated on the platform in October 2022, but the underlying data originates from 2015.

Geography

Sinhala language suggests primary relevance to Sri Lanka.