# EduChat: A Large-Scale Language Model-based Chatbot System for Intelligent Education

**Authors:** Yuhao Dan, Zhikai Lei, Yiyang Gu, Yong Li, Jianghao Yin, Jiaju Lin, Linhao Ye, Zhiyan Tie, Yougen Zhou, Yilei Wang, Aimin Zhou, Ze Zhou, Qin Chen, Jie Zhou, Liang He, Xipeng Qiu  
**arXiv:** [2308.02773](https://arxiv.org/abs/2308.02773)  
**Date:** 2023-08-05  
**Summary page:** https://www.summarizepaper.com/en/paper/2308.02773v1/

## Key points

- EduChat is a large-scale language model (LLM)-based chatbot system designed for intelligent education.
- It provides personalized, fair, and compassionate support to teachers, students, and parents.
- Offers various educational functions such as open question answering, essay assessment, Socratic teaching, and emotional support.
- Uses real-time updated corpora from the internet as an external knowledge source for open question answering.
- Provides fine-grained and comprehensive feedback for essay assessment.
- Focuses on developing Socratic teaching capabilities to stimulate discussions and foster advanced cognitive skills.
- Incorporates a psychology-based emotional support framework to provide personalized diagnoses and emotional support tailored to individual users' needs.
- Training procedure involves two stages: fundamental capabilities acquisition and educational skills acquisition.

## Summary

EduChat is a large-scale language model (LLM)-based chatbot system designed for intelligent education. It provides personalized, fair, and compassionate support to teachers, students, and parents through various educational functions such as open question answering, essay assessment, Socratic teaching, and emotional support. One of its key features is the use of real-time updated corpora from the internet as an external knowledge source for open question answering. This ensures accurate and up-to-date information in responses. In terms of essay assessment, EduChat goes beyond coarse grading by offering fine-grained and comprehensive feedback. It combines the expertise of teaching professionals to provide overall scores, aspect-level ratings, and detailed comments on content, expression, paragraph structure, and overall evaluation. This allows for personalized guidance in all aspects of writing. EduChat also focuses on developing Socratic teaching capabilities rather than providing direct answers to students. Through multi-step question-and-answer interactions, it stimulates discussions, debates, evaluations,and analyses to foster advanced cognitive skills and cultivate students' autonomy in learning. Recognizing the psychological pressures faced by adolescents and children,EduChat incorporates a psychology-based emotional support framework. By simulating a psychological counselor based on emotion psychology theories like Rational Emotive Behavior Therapy (REBT)and the ABC theory (Ellis), EduChat provides personalized diagnoses and emotional support tailored to individual users' needs. The training procedure of EduChat involves two stages: fundamental capabilities acquisition and educational skills acquisition.In the first stage,the model is pre-trained on educational books and Q&A pairs to establish foundational knowledge across disciplines.Large-scale instruction tuningand open-domain dialogue datasets are also incorporated to enable basic instruction following abilityand dialogue ability. In the second stage,EduChat further hones its abilities through domain-specific training in educational skills.This ensures that it can effectively fulfill its role as an intelligent education tool,providing a comprehensive and personalized learning experience for users.