Saved in:
Bibliographic Details
Main Authors: Fan, Zhilin, Wang, Deliang, Chen, Penghe, Lu, Yu
Format: Preprint
Published: 2026
Subjects:
Online Access:https://arxiv.org/abs/2604.22237
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866910162005000192
author Fan, Zhilin
Wang, Deliang
Chen, Penghe
Lu, Yu
author_facet Fan, Zhilin
Wang, Deliang
Chen, Penghe
Lu, Yu
contents Diagnosing student problem behaviors requires teachers to synthesize multifaceted information, identify behavioral categories, and plan intervention strategies. Although fine-tuned large language models (LLMs) can support this process through multi-turn dialogue, they rarely explain why a strategy is recommended, limiting transparency and teachers' trust. To address this issue, we present an explainable dialogue system built on a fine-tuned LLM. The system uses a hierarchical attribution method based on explainable AI (xAI) to identify dialogue evidence for each recommendation and generate a natural-language explanation based on that evidence. In technical evaluation, the method outperformed baseline approaches in identifying supporting evidence. In a preliminary user study with 22 pre-service teachers, participants who received explanations reported higher trust in the system. These findings suggest a promising direction for improving LLM explainability in educational dialogue systems.
format Preprint
id arxiv_https___arxiv_org_abs_2604_22237
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Tell Me Why: Designing an Explainable LLM-based Dialogue System for Student Problem Behavior Diagnosis
Fan, Zhilin
Wang, Deliang
Chen, Penghe
Lu, Yu
Computation and Language
Artificial Intelligence
Diagnosing student problem behaviors requires teachers to synthesize multifaceted information, identify behavioral categories, and plan intervention strategies. Although fine-tuned large language models (LLMs) can support this process through multi-turn dialogue, they rarely explain why a strategy is recommended, limiting transparency and teachers' trust. To address this issue, we present an explainable dialogue system built on a fine-tuned LLM. The system uses a hierarchical attribution method based on explainable AI (xAI) to identify dialogue evidence for each recommendation and generate a natural-language explanation based on that evidence. In technical evaluation, the method outperformed baseline approaches in identifying supporting evidence. In a preliminary user study with 22 pre-service teachers, participants who received explanations reported higher trust in the system. These findings suggest a promising direction for improving LLM explainability in educational dialogue systems.
title Tell Me Why: Designing an Explainable LLM-based Dialogue System for Student Problem Behavior Diagnosis
topic Computation and Language
Artificial Intelligence
url https://arxiv.org/abs/2604.22237