Scholarlink

<p>Accurate estimation or evaluation of English translations remains a significant challenge for Chinese–English tasks due to difficulties in capturing contextual meaning, syntactic correctness, and learner-oriented feedback. The primary goal is to design a machine-assisted translation quality evaluation framework that improves assessment accuracy, consistency, and interpretability while supporting translation learning and feedback. The proposed approach integrates automated semantic and syntactic analysis with interactive human feedback to deliver a comprehensive and learner-centered evaluation of translation quality. The framework assesses fluency, grammatical correctness, and semantic alignment between source and translated texts. The framework is evaluated using academic Chinese–English translation tasks collected from Chinese university learning platforms, including Wenjuanxing, WeLearn, and Xuexitong, together with publicly available parallel corpora. Experimental results demonstrate strong evaluation performance, showing consistently higher translation quality assessment accuracy and reliability compared with existing evaluation approaches. The main contributions of this study include the development of a unified translation evaluation framework that combines automated analysis with human feedback, empirical validation on real academic translation data, and evidence of improved evaluation effectiveness. The proposed framework offers practical value for automated translation assessment and translation pedagogy in higher education.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

A machine-assisted framework for english translation quality evaluation based on deep learning and multi-view graph convolution

  • 期刊论文

  • 发表日期：

    2026年6月22日

查看全文
  • Ting Duan

摘要

Accurate estimation or evaluation of English translations remains a significant challenge for Chinese–English tasks due to difficulties in capturing contextual meaning, syntactic correctness, and learner-oriented feedback. The primary goal is to design a machine-assisted translation quality evaluation framework that improves assessment accuracy, consistency, and interpretability while supporting translation learning and feedback. The proposed approach integrates automated semantic and syntactic analysis with interactive human feedback to deliver a comprehensive and learner-centered evaluation of translation quality. The framework assesses fluency, grammatical correctness, and semantic alignment between source and translated texts. The framework is evaluated using academic Chinese–English translation tasks collected from Chinese university learning platforms, including Wenjuanxing, WeLearn, and Xuexitong, together with publicly available parallel corpora. Experimental results demonstrate strong evaluation performance, showing consistently higher translation quality assessment accuracy and reliability compared with existing evaluation approaches. The main contributions of this study include the development of a unified translation evaluation framework that combines automated analysis with human feedback, empirical validation on real academic translation data, and evidence of improved evaluation effectiveness. The proposed framework offers practical value for automated translation assessment and translation pedagogy in higher education.