矿业工程专业专门用途英语大模型智能体“Mining Lingua”的研发及其对教学的启示

Developing an ESP-Oriented LLM Agent “Mining Lingua” for Mining Engineering and Its Pedagogical Implications

摘要: 在高校工程类专业课程体系中，专门用途英语教学实践常存在课堂教学内容与学习者自适应学习、实际工作场景交流需求脱节的情况。大模型智能体的研发有助于实现第一课堂与第二课堂学习成果的衔接与过渡，推动从课堂到工作场景的“智能拓学”。本研究以矿业工程专业为例，针对该专业专门用途英语常见的学术科研与工程管理场景，分别构建学术英语与工程管理英语大型语料库作为知识库，搭建专门用途英语大模型智能体“Mining Lingua”。该智能体具备术语解析、语言润色、中英互译、句型与模板生成、概念溯源等5个主要功能，可应用在基于体裁分析的显性教学、成果导向教育(OBE)教学及人机协同教学当中，为学习者提供个性化、场景化、交互式的语言学习支持，有效提升学习者的文体与语体意识及人机协同素养。

Abstract: In the current curriculum of engineering disciplines, ESP pedagogical practices have been limited for the discontinuity between teaching content in classroom settings, learners’ adaptive learning, and their communicative needs in actual working scenarios. In this light, the development of large language model (LLM) agents facilitates the bridging of learning outcomes between formal classroom instruction and informal out-of-class learning, promoting “intelligent extended learning” from academic settings to professional workplaces. Taking mining engineering as a case study, this study constructs large-scale corpora as knowledge bases for two typical ESP scenarios—academic research and engineering management—and develops an LLM-powered, ESP-oriented intelligent agent named “Mining Lingua”. The agent features five core functions: analyzing terminologies, polishing language, C-E/E-C translation, generating syntactic patterns/textual templates, and tracing conceptual terms. In ESP teaching, this LLM agent could be applied in genre-based explicit instructions, OBE teaching, and human-machine collaborative teaching. It could offer learners personalized, contextualized, and interactive language-learning support, enhance their genre and register awareness, and promote their digital literacy in human-machine collaborative learning.