ISHIDA International

This data product is an evaluation sample of a French–Japanese legal parallel corpus curated from professionally translated legal documents. Scope: The dataset focuses on sentence-level French–Japanese parallel text from legal and contractual materials, including corporate, commercial, and regulatory language. All sentence pairs are manually aligned and normalized for linguistic consistency.

Scale: This sample contains approximately 13,000 aligned sentence pairs and is provided as a subset of a larger proprietary legal corpus currently under expansion. The full production dataset is planned to exceed 20,000 sentence pairs. Value: The dataset is designed for evaluation and benchmarking purposes, enabling organizations to assess translation quality, sentence alignment accuracy, and domain-specific linguistic consistency.

Typical use cases include machine translation evaluation, legal NLP benchmarking, and controlled LLM assessment. The dataset does not contain personal data and is not intended for large-scale model training. It is provided as a high-quality evaluation sample to support research, testing, and comparative analysis workflows.