Fincorpus: Chinese Financial News, Corporate Announcements, and Exam Questions

A 2023 collection of Chinese financial text data from Duxiaoman-DI. It includes corporate announcements, financial news/articles, and financial exam questions, with each record containing a text field and a source metadata field. The total uncompressed size of the components is approximately 60.37 gigabytes.

Use Cases

• Train financial domain language models based on the large corpus of news and articles.
• Analyze corporate disclosure patterns based on the collection of listed company announcements.
• Develop or benchmark financial question-answering systems based on the included exam questions.
• Perform financial sentiment analysis or event extraction based on the news and announcement text.

Strengths

• Large-scale text corpus with approximately 60.37 GB of uncompressed data.
• Contains multiple distinct data types: announcements, news/articles, and exam questions.
• Includes source metadata for each text entry, allowing for provenance tracking.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Last updated 2023-09-22 10:10:10; freshness should be verified.
• Row count is unknown, which may limit suitability assessment.

Provenance

Source

Duxiaoman-DI

Collection Method

Aggregated from unspecified Chinese financial sources.

Freshness

Last updated 2023-09-22 10:10:10.

Geography

China (implied by language and content)