@inproceedings{nehrdich-etal-2025-mitra,
    title = "{MITRA}-zh-eval: Using a Buddhist {C}hinese Language Evaluation Dataset to Assess Machine Translation and Evaluation Metrics",
    author = "Nehrdich, Sebastian  and
      Chen, Avery  and
      Bingenheimer, Marcus  and
      Huang, Lu  and
      Tang, Rouying  and
      Wei, Xiang  and
      Zhu, Leijie  and
      Keutzer, Kurt",
    editor = {H{\"a}m{\"a}l{\"a}inen, Mika  and
      {\"O}hman, Emily  and
      Bizzoni, Yuri  and
      Miyagawa, So  and
      Alnajjar, Khalid},
    booktitle = "Proceedings of the 5th International Conference on Natural Language Processing for Digital Humanities",
    month = may,
    year = "2025",
    address = "Albuquerque, USA",
    publisher = "Association for Computational Linguistics",
    url = "https://preview.aclanthology.org/fix-sig-urls/2025.nlp4dh-1.12/",
    pages = "129--137",
    ISBN = "979-8-89176-234-3",
    abstract = "With the advent of large language models, machine translation (MT) has become a widely used, but little understood, tool for accessing historical and multilingual texts. While models like GPT, Claude, and Deepseek increasingly enable translation of low-resource and ancient languages, critical questions remain about their evaluation, optimal model selection, and the value of domain-specific training and retrieval-augmented generation setups.While AI models like GPT, Claude, and Deepseek are improving translation capabilities for low-resource and ancient languages, researchers still face important questions about how to evaluate their performance, which models work best, and whether specialized training approaches provide meaningful improvements in translation quality.This study introduces a comprehensive evaluation dataset for Buddhist Chinese to English translation, comprising 2,662 bilingual data points from 32 texts that have been selected to represent the full breadth of the Chinese Buddhist canon.We evaluate various computational metrics of translation quality (BLEU, chrF, BLEURT, GEMBA) against expert annotations from five domain specialists who rated 182 machine-generated translations. Our analysis reveals that LLM-based GEMBA scoring shows the strongest correlation with human judgment, significantly outperforming traditional metrics. We then benchmark commercial models (GPT-4 Turbo, Claude 3.5, Gemini), open-source models (Gemma 2, Deepseek-r1), and a domain-specialized model (Gemma 2 Mitra) using GEMBA. Our results demonstrate that domain-specific training enables open-weights models to achieve competitive performance with commercial systems, while also showing that retrieval-augmented generation (RAG) significantly improves translation quality for the best performing commercial models."
}