Japanese Kana-to-Kanji Conversion Dataset with Context for Disambiguation

A Japanese kana-to-kanji conversion dataset with context for disambiguation, containing 77,277,970 entries. The dataset was created by author katsukiono and was last updated on Hugging Face on 2026-01 08. Data is provided in JSONL format with entries containing an input context string and output kanji candidates.

Use Cases

• Training kana-to-kanji conversion models based on the context and reading data.
• Evaluating disambiguation algorithms for Japanese text based on the multiple candidate outputs.
• Building or benchmarking Japanese input method editors (IMEs) based on the large-scale conversion examples.

Strengths

• Large scale with over 77 million entries.
• Data format includes context for disambiguation, as shown in the examples.

Limitations

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

Provenance

Source

huggingface

Collection Method

Likely compiled from text sources for conversion tasks.

Time Range

null

Freshness

Last updated 2026-01-08 02:47:36; freshness should be verified.

Geography

null