Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Curriculum-Driven Edubot: A Framework for Developing Language Learning Chatbots Through Synthesizing Conversational Data

Yu Li, Shang Qu, Jili Shen, Shangchao Min, Zhou Yu

TL;DR

This work introduces Curriculum-Driven EduBot, a chatbot framework that aligns conversational practice with textbook curricula to support language learning. It synthesizes curriculum-based dialogues by augmenting topics, creating fixed-format personas, and incorporating textbook vocabulary, then fine-tunes an open-source model (Vicuna-13B) using those dialogues. A user study shows EduBot outperforms ChatGPT in curriculum-aligned dialogue and proficiency-adaptive interaction, while delivering more natural, persona-aware conversations. The framework offers a scalable approach to integrating structured curriculum content with interactive AI, potentially enhancing targeted language learning outcomes across curricula.

Abstract

Chatbots have become popular in educational settings, revolutionizing how students interact with material and how teachers teach. We present Curriculum-Driven EduBot, a framework for developing a chatbot that combines the interactive features of chatbots with the systematic material of English textbooks to assist students in enhancing their conversational skills. We begin by extracting pertinent topics from textbooks and using large language models to generate dialogues related to these topics. We then fine-tune an open-source model using our generated conversational data to create our curriculum-driven chatbot. User studies demonstrate that EduBot outperforms ChatGPT in leading curriculum-based dialogues and adapting its dialogue to match the user's English proficiency level. By combining traditional textbook methodologies with conversational AI, our approach offers learners an interactive tool that aligns with their curriculum and provides user-tailored conversation practice. This facilitates meaningful student-bot dialogues and enriches the overall learning experience within the curriculum's pedagogical framework.

Curriculum-Driven Edubot: A Framework for Developing Language Learning Chatbots Through Synthesizing Conversational Data

TL;DR

This work introduces Curriculum-Driven EduBot, a chatbot framework that aligns conversational practice with textbook curricula to support language learning. It synthesizes curriculum-based dialogues by augmenting topics, creating fixed-format personas, and incorporating textbook vocabulary, then fine-tunes an open-source model (Vicuna-13B) using those dialogues. A user study shows EduBot outperforms ChatGPT in curriculum-aligned dialogue and proficiency-adaptive interaction, while delivering more natural, persona-aware conversations. The framework offers a scalable approach to integrating structured curriculum content with interactive AI, potentially enhancing targeted language learning outcomes across curricula.

Abstract

Chatbots have become popular in educational settings, revolutionizing how students interact with material and how teachers teach. We present Curriculum-Driven EduBot, a framework for developing a chatbot that combines the interactive features of chatbots with the systematic material of English textbooks to assist students in enhancing their conversational skills. We begin by extracting pertinent topics from textbooks and using large language models to generate dialogues related to these topics. We then fine-tune an open-source model using our generated conversational data to create our curriculum-driven chatbot. User studies demonstrate that EduBot outperforms ChatGPT in leading curriculum-based dialogues and adapting its dialogue to match the user's English proficiency level. By combining traditional textbook methodologies with conversational AI, our approach offers learners an interactive tool that aligns with their curriculum and provides user-tailored conversation practice. This facilitates meaningful student-bot dialogues and enriches the overall learning experience within the curriculum's pedagogical framework.
Paper Structure (55 sections, 20 figures, 3 tables)

This paper contains 55 sections, 20 figures, 3 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. Method
  4. Conversational Data Augmentation
  5. Augment Topics
  6. Create Personas
  7. Compose Dialogues
  8. Fine-Tuning An Open-Source Language Model With Synthesized Conversational Data
  9. Deployment of the Fine-tuned EduBot
  10. Experiments
  11. Curriculum Source
  12. Baseline
  13. Experimental Settings
  14. Participants
  15. Procedures
  16. ...and 40 more sections

Figures (20)

  • Figure 1: Comparison between ChatGPT vs. Our Curriculum-Driven Edubot. ChatGPT operates as an AI-powered Q&A tool, delivering comprehensive responses from a broad knowledge base. The Curriculum-Driven Edubot is fine-tuned with synthesized conversations, offering an interactive and adaptive learning experience through conversational practice.
  • Figure 2: The initial step of the Curriculum-Driven EduBot Development is to enhance textbook topics (Sec.\ref{['augment_topics']}). Following this, personas are created for synthetic conversation participants (Sec.\ref{['create_personas']}). Then dialogues are constructed based on vocabulary, topics and personas (Sec. \ref{['compose_dialogues']}). After this, an open-source model Vicuna is fine-tuned to get the EduBot ready for deployment (Sec. \ref{['fine_tune_model']})
  • Figure 3: Case 1
  • Figure 4: Case 2 from EduBot and Case 3 from ChatGPT-Bot.
  • Figure 5: Case 4 and Case 5.
  • ...and 15 more figures