agentlans/en-zhtw-google-translate

English-Traditional Chinese Translation Dataset

A parallel dataset of 1,000,000 high-quality English and Traditional Chinese (Taiwan) sentences, machine-translated using Google Translate.

This dataset combines sentences from two curated sources:

• agentlans/Taiwan-Text-Excellence-sentences (Traditional Chinese)
• agentlans/high-quality-english-sentences (English)

Each sentence from one source was translated to the other language via Google Translate, creating bidirectional pairs. The original field indicates the source language.

Dataset Structure

Split	Examples	% of Total
train	799,616	80.0%
test	199,905	20.0%

Total examples: 999,521

Example Row

{
  "en": "There is also a rumor about the Sibu Xiang. It is said that anyone who encounters the Sibu Xiang will encounter bad luck, such as failing exams, repeating a grade, dropping out of school, and even their university life and life will be unsuccessful.",
  "zhtw": "關於四不像亦有一則傳聞，相傳碰到四不像的人都會遭遇厄運，如考試不合格、重讀、退學，連大學生活和人也亦會事事不順。",
  "original": "zhtw"
}

Fields:

• en: English sentence
• zhtw: Traditional Chinese (Taiwan) sentence
• original: Source language ("en" or "zhtw")

Dataset Creation

1. Selected 1,000,000 sentences from each source dataset
2. Translated English → Traditional Chinese (500k pairs)
3. Translated Traditional Chinese → English (500k pairs)
4. Split 80/20 into train/test sets

Limitations

• Single sentences lack discourse-level context which limits quality and smoothness of the translations
• Inconsistent Romanization/romanized terms across translations
• Structural and stylistic differences between English and Chinese, especially idioms and culture-specific terms
• Machine translation quality (Google Translate)

Licence

• Taiwan-Text-Excellence-sentences: CC-BY-NC-SA 4.0
• high-quality-english-sentences: ODC-BY