Mandarin Business Translation Evaluation Dataset (AI vs. Human)

This dataset contains a comparative evaluation corpus used to analyze AI-generated translations of Mandarin business texts against human reference translations. The data were collected as part of a study assessing the semantic fidelity of machine translation models, including ChatGPT, Google Translate, and Microsoft Copilot, within formal business communication contexts. Each record consists of parallel text pairs comprising the original Mandarin source sentence, the human reference translation, and outputs generated by the three AI systems. Additionally, the dataset includes semantic similarity scores computed using embedding-based models such as Sentence-BERT (SBERT), Universal Sentence Encoder (USE), and XLM-RoBERTa (XLM-R).