BROWSE
Detail
Conference Paper
Developing Conversational Intelligent Tutoring for Speaking Skills in Second Language Learning
Cited 0 time in 
Share 
Share
- Authors
- Jeongmin Lee, Jin-Xia Huang, Minsoo Cho, Yoon-Hyung Roh, Oh-Woog Kwon, Yunkeun Lee
- Issue Date
- 2024-06
- Citation
- International Conference on Intelligent Tutoring Systems (ITS) 2024 (LNCS 14798), pp.131-148
- Publisher
- Springer
- Language
- English
- Type
- Conference Paper
- Abstract
- In this study, we introduce a Conversational Intelligent Tutoring System designed to create an interactive and immersive learning experience aimed at improving the speaking abilities of second language learners. This system mimics a human tutor by engaging in role-play dialogues with the learner, based on predefined scenarios, and offers corrective feedback on the learners’ utterance, while also engaging in chat to encourage student participation. The tutoring system includes a deep-learning classifier to assess students’ utterances, a dialogue generator customized for responding to students’ free-form chats, and a straightforward dialogue manager to determine the sequence of conversational turns. To tackle the challenge posed by the scarcity of tutoring dialogue resources, a significant hurdle for deep learning methods, we present a cost-effective approach that efficiently extends existing Korean dialogue datasets for the purpose of intelligent language tutoring. We carried out a series of experiments to compare various fine-tuned models based on language models of different sizes, and included a comparative analysis with ChatGPT. We discovered that smaller, specialized, fine-tuned models can either surpass or match the performance of GPT-4 in specific tutoring applications. Given the educational sector’s demand for cost-effective solutions, our contributions, spanning system design, dataset development, and comparative analysis, serve as valuable references to address these needs.
- KSP Keywords
- Comparative analysis, Corrective feedback, Cost-effective approach, Dialogue Manager, Different sizes, Free-form, Intelligent tutoring System, Language Model, Learning Experience, Learning methods, Role-play