Lian, W., Zheng, M., & Xu, J. (2024). A Comparative Study on Development Models of AI Chinese Language Partners Based on the ERNIE Large Language Model. Journal of Technology and Chinese Language Teaching, 15(2), 35-53.
[连维琛, 郑明鉴, & 徐娟. (2024). 基于文心大模型的AI中文语伴开发模式对比研究. 科技与中文教学 (Journal of Technology and Chinese Language Teaching), 15(2), 35-53.]

Full paper

Abstract/摘要：

The importance of language partners in language learning is undeniable. In the context of the rapid development of generative artificial intelligence technology, this study attempts to construct an AI Chinese language partner based on large language models, utilizing the ERNIE model as the foundational model. By crafting prompts and integrating local Chinese teaching resources, the general model is tailored to better suit the field of international Chinese education. This research develops AI Chinese language partners for Chinese learners using both low-code and code-based development modes, comparing the development processes of these two approaches. Furthermore, the study analyzes the developed AI language partners in terms of basic functionalities and generated content, providing recommendations for their selection and use.

语伴在语言学习中的重要性不言而喻，在生成式人工智能技术飞速发展的时代背景下，本研究尝试依托大语言模型构建AI中文语伴，选用百度文心一言大模型（ERNIE）作为基础大模型，通过编写提示词（Prompt）、集成本地中文教学资源等方式使通用大模型更适用于国际中文教育领域。本研究分别用低码态与代码态两种开发模式开发面向中文学习者的AI中文语伴，并对两种开发模式的开发过程进行对比，然后从基本功能、生成内容等方面对开发的AI中文语伴进行分析并给出择用建议。