PatSnap Translation: Chinese-English Patent Parallel Corpus

268 full-text patents constitute this Chinese-English parallel corpus, sourced from public patent specifications. Each entry has been verified by human experts for compliance with patent translation and drafting standards. The dataset was created by PatSnap and forms part of the training data for Hiro-Translation.

Use Cases

• Train machine translation models based on the high-quality Chinese-English parallel corpus.
• Benchmark translation systems on technical patent language based on the verified patent specifications.
• Fine-tune language models for domain-specific translation tasks based on the patent drafting standards.

Strengths

• 268 full-text patents provide a focused corpus for technical translation.
• Human expert verification ensures compliance with patent translation and drafting standards.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count and total size are unknown, which may limit suitability assessment.

Provenance

Source

PatSnap

Collection Method

Sourced from publicly available patent specifications.

Freshness

Last updated 2026-06-09 01:51:31; freshness should be verified.