Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Personality-aware Student Simulation for Conversational Intelligent Tutoring Systems

Zhengyuan Liu, Stella Xin Yin, Geyu Lin, Nancy F. Chen

TL;DR

This work addresses the challenge of characterizing and simulating student persona in conversational ITSs by introducing a personality-aware framework that integrates cognitive (language ability) and noncognitive (BF-TC) traits. It refines the Big Five into BF-TC for tutoring dialogue and employs multi-aspect validation to ensure alignment with pedagogical goals, language ability, and standardized personality measures. Through a language-learning case study, the authors demonstrate that state-of-the-art LLMs, especially GPT-4, can generate diverse student responses and adapt teacher scaffolding according to personality and ability, with strong consistency to Vanilla BFI metrics. The framework advances scalable evaluation and design of personalized conversational ITSs, enabling more engaging, self-paced, and tailored learning experiences, while noting limitations such as language scope and model biases. This work lays a foundation for robust, persona-aware tutoring systems across domains and languages, with implications for pedagogy, assessment, and human-AI collaboration in education.

Abstract

Intelligent Tutoring Systems (ITSs) can provide personalized and self-paced learning experience. The emergence of large language models (LLMs) further enables better human-machine interaction, and facilitates the development of conversational ITSs in various disciplines such as math and language learning. In dialogic teaching, recognizing and adapting to individual characteristics can significantly enhance student engagement and learning efficiency. However, characterizing and simulating student's persona remain challenging in training and evaluating conversational ITSs. In this work, we propose a framework to construct profiles of different student groups by refining and integrating both cognitive and noncognitive aspects, and leverage LLMs for personality-aware student simulation in a language learning scenario. We further enhance the framework with multi-aspect validation, and conduct extensive analysis from both teacher and student perspectives. Our experimental results show that state-of-the-art LLMs can produce diverse student responses according to the given language ability and personality traits, and trigger teacher's adaptive scaffolding strategies.

Personality-aware Student Simulation for Conversational Intelligent Tutoring Systems

TL;DR

This work addresses the challenge of characterizing and simulating student persona in conversational ITSs by introducing a personality-aware framework that integrates cognitive (language ability) and noncognitive (BF-TC) traits. It refines the Big Five into BF-TC for tutoring dialogue and employs multi-aspect validation to ensure alignment with pedagogical goals, language ability, and standardized personality measures. Through a language-learning case study, the authors demonstrate that state-of-the-art LLMs, especially GPT-4, can generate diverse student responses and adapt teacher scaffolding according to personality and ability, with strong consistency to Vanilla BFI metrics. The framework advances scalable evaluation and design of personalized conversational ITSs, enabling more engaging, self-paced, and tailored learning experiences, while noting limitations such as language scope and model biases. This work lays a foundation for robust, persona-aware tutoring systems across domains and languages, with implications for pedagogy, assessment, and human-AI collaboration in education.

Abstract

Intelligent Tutoring Systems (ITSs) can provide personalized and self-paced learning experience. The emergence of large language models (LLMs) further enables better human-machine interaction, and facilitates the development of conversational ITSs in various disciplines such as math and language learning. In dialogic teaching, recognizing and adapting to individual characteristics can significantly enhance student engagement and learning efficiency. However, characterizing and simulating student's persona remain challenging in training and evaluating conversational ITSs. In this work, we propose a framework to construct profiles of different student groups by refining and integrating both cognitive and noncognitive aspects, and leverage LLMs for personality-aware student simulation in a language learning scenario. We further enhance the framework with multi-aspect validation, and conduct extensive analysis from both teacher and student perspectives. Our experimental results show that state-of-the-art LLMs can produce diverse student responses according to the given language ability and personality traits, and trigger teacher's adaptive scaffolding strategies.
Paper Structure (20 sections, 5 figures, 9 tables)

This paper contains 20 sections, 5 figures, 9 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. Intelligent Tutoring Systems
  4. Personality in Education & Language Learning
  5. User Simulation for Dialogue Systems
  6. Personality-aware Student Simulation & Multi-aspect Validation Framework
  7. Cognitive Level Simulation
  8. Noncognitive Level Simulation
  9. Multi-aspect Validation Framework
  10. Our Experiments on Language Tutoring Conversations
  11. Task Description & Role Setting
  12. Experimental Setup
  13. Results and Analysis
  14. Effectiveness of BF-TC Simulation
  15. Consistency between BF-TC and BFI
  16. ...and 5 more sections

Figures (5)

  • Figure 1: Tutoring conversation segments of two students with different personality traits.
  • Figure 2: Overview of our proposed framework for personality-aware simulation and multi-aspect validation.
  • Figure 3: Student response embedding distribution of simulation w/o BF-TC (blue) and w/ BF-TC (orange).
  • Figure 4: Heatmap of the correlation between personality traits and scaffolding strategies. Left: students with high language ability. Right: students with low language ability. Experimented Model: GPT-4-1106.
  • Figure 5: Correlation between language ability and scaffolding categorization. p values are <.05