Thucnews Chinese News Corpus for NER

Thucnews is a Chinese-language news dataset from Tsinghua University, primarily designed for named entity recognition tasks. The dataset was created by 'seamew' and last updated on the Hugging Face platform in June 2021.

Use Cases

• Train named entity recognition models to identify person, location, and organization names within Chinese news text.
• Benchmark text classification algorithms on news article categorization using labels inferred from the dataset's purpose.
• Analyze linguistic patterns and entity co-occurrence within a corpus of contemporary Chinese news writing.

Strengths

• Dataset originates from Tsinghua University, a leading academic institution.
• Specifically curated for the well-defined task of named entity recognition in Chinese.

Limitations

• The exact number of rows, articles, and entities is unknown, limiting reproducibility assessments.
• Data is from 2021 or earlier, potentially not reflecting current language or news topics.
• Potential geographic or topical bias inherent in the original news source collection is unquantified.

Provenance

Source

Tsinghua University (THUC).

Collection Method

Collection of Chinese news articles, method details unknown.

Time Range

null

Freshness

Last updated in 2021; static dataset with no stated update frequency.

Geography

Presumably China, given the language and source.