Guardado en:
Detalles Bibliográficos
Autores principales: Zhao, Xiying, Wen, Zhoufutu, Chen, Zhixuan, Ding, Jingzhe, Jiao, Jianpeng, Li, Shuai, Li, Xi, Liang, Danni, Long, Shengda, Liu, Qianqian, Wu, Xianbo, Gao, Hongwan, Gao, Xiang, Hu, Liang, Liu, Jiashuo, Liu, Mengyun, Shi, Weiran, Yang, Chenghao, Yang, Qianyu, Zhang, Xuanliang, Zhang, Ge, Huang, Wenhao, Tang, Yuwen
Formato: Preprint
Publicado: 2025
Materias:
Acceso en línea:https://arxiv.org/abs/2511.10984
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866909965957988352
author Zhao, Xiying
Wen, Zhoufutu
Chen, Zhixuan
Ding, Jingzhe
Jiao, Jianpeng
Li, Shuai
Li, Xi
Liang, Danni
Long, Shengda
Liu, Qianqian
Wu, Xianbo
Gao, Hongwan
Gao, Xiang
Hu, Liang
Liu, Jiashuo
Liu, Mengyun
Shi, Weiran
Yang, Chenghao
Yang, Qianyu
Zhang, Xuanliang
Zhang, Ge
Huang, Wenhao
Tang, Yuwen
author_facet Zhao, Xiying
Wen, Zhoufutu
Chen, Zhixuan
Ding, Jingzhe
Jiao, Jianpeng
Li, Shuai
Li, Xi
Liang, Danni
Long, Shengda
Liu, Qianqian
Wu, Xianbo
Gao, Hongwan
Gao, Xiang
Hu, Liang
Liu, Jiashuo
Liu, Mengyun
Shi, Weiran
Yang, Chenghao
Yang, Qianyu
Zhang, Xuanliang
Zhang, Ge
Huang, Wenhao
Tang, Yuwen
contents The evaluation of discourse-level translation in expert domains remains inadequate, despite its centrality to knowledge dissemination and cross-lingual scholarly communication. While these translations demand discourse-level coherence and strict terminological precision, current evaluation methods predominantly focus on segment-level accuracy and fluency. To address this limitation, we introduce DiscoX, a new benchmark for discourse-level and expert-level Chinese-English translation. It comprises 200 professionally-curated texts from 7 domains, with an average length exceeding 1700 tokens. To evaluate performance on DiscoX, we also develop Metric-S, a reference-free system that provides fine-grained automatic assessments across accuracy, fluency, and appropriateness. Metric-S demonstrates strong consistency with human judgments, significantly outperforming existing metrics. Our experiments reveal a remarkable performance gap: even the most advanced LLMs still trail human experts on these tasks. This finding validates the difficulty of DiscoX and underscores the challenges that remain in achieving professional-grade machine translation. The proposed benchmark and evaluation system provide a robust framework for more rigorous evaluation, facilitating future advancements in LLM-based translation.
format Preprint
id arxiv_https___arxiv_org_abs_2511_10984
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle DiscoX: Benchmarking Discourse-Level Translation task in Expert Domains
Zhao, Xiying
Wen, Zhoufutu
Chen, Zhixuan
Ding, Jingzhe
Jiao, Jianpeng
Li, Shuai
Li, Xi
Liang, Danni
Long, Shengda
Liu, Qianqian
Wu, Xianbo
Gao, Hongwan
Gao, Xiang
Hu, Liang
Liu, Jiashuo
Liu, Mengyun
Shi, Weiran
Yang, Chenghao
Yang, Qianyu
Zhang, Xuanliang
Zhang, Ge
Huang, Wenhao
Tang, Yuwen
Computation and Language
Artificial Intelligence
The evaluation of discourse-level translation in expert domains remains inadequate, despite its centrality to knowledge dissemination and cross-lingual scholarly communication. While these translations demand discourse-level coherence and strict terminological precision, current evaluation methods predominantly focus on segment-level accuracy and fluency. To address this limitation, we introduce DiscoX, a new benchmark for discourse-level and expert-level Chinese-English translation. It comprises 200 professionally-curated texts from 7 domains, with an average length exceeding 1700 tokens. To evaluate performance on DiscoX, we also develop Metric-S, a reference-free system that provides fine-grained automatic assessments across accuracy, fluency, and appropriateness. Metric-S demonstrates strong consistency with human judgments, significantly outperforming existing metrics. Our experiments reveal a remarkable performance gap: even the most advanced LLMs still trail human experts on these tasks. This finding validates the difficulty of DiscoX and underscores the challenges that remain in achieving professional-grade machine translation. The proposed benchmark and evaluation system provide a robust framework for more rigorous evaluation, facilitating future advancements in LLM-based translation.
title DiscoX: Benchmarking Discourse-Level Translation task in Expert Domains
topic Computation and Language
Artificial Intelligence
url https://arxiv.org/abs/2511.10984