Chinese Patent Records from 2003 to 2020, Approximately 18 Million Documents

TMSDMAP's Chinese Patent Dataset contains approximately 18 million patent records published in China between 2003 and 2020. The dataset includes invention patents and utility models, and is described as a high-quality corpus for research and model training. The original data format is JSON, with an overall size of about 250 GB.

Use Cases

• Analyzing technology evolution and innovation trends in China based on patent publication data.
• Training legal or domain-specific language models based on the high-quality patent text corpus.
• Studying intellectual property protection patterns and patent types over time.
• Conducting temporal analysis of patenting activity across a 17-year period.

Strengths

• Approximately 18 million records, providing substantial scale for analysis.
• Covers a 17-year time range from 2003 to 2020, enabling longitudinal study.
• Includes multiple patent types such as invention patents and utility models.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is an approximation; exact record count is unknown.
• Description metadata is limited; actual data quality requires manual inspection after download.

Provenance

Source

TMSDMAP

Collection Method

Likely compiled from official Chinese patent publication sources.

Time Range

2003-2020

Freshness

Last updated 2026-04-06 11:04:14; freshness should be verified.

Geography

China