Saved in:
Bibliographic Details
Main Authors: Pei, Jiahuan, Ye, Fanghua, Sun, Xin, Deng, Wentao, Hindriks, Koen, Wang, Junxiao
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2507.05528
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866908520887091200
author Pei, Jiahuan
Ye, Fanghua
Sun, Xin
Deng, Wentao
Hindriks, Koen
Wang, Junxiao
author_facet Pei, Jiahuan
Ye, Fanghua
Sun, Xin
Deng, Wentao
Hindriks, Koen
Wang, Junxiao
contents Large language models (LLMs) have advanced virtual educators and learners, bridging NLP with AI4Education. Existing work often lacks scalability and fails to leverage diverse, large-scale course content, with limited frameworks for assessing pedagogic quality. To this end, we propose WikiHowAgent, a multi-agent workflow leveraging LLMs to simulate interactive teaching-learning conversations. It integrates teacher and learner agents, an interaction manager, and an evaluator to facilitate procedural learning and assess pedagogic quality. We introduce a dataset of 114,296 teacher-learner conversations grounded in 14,287 tutorials across 17 domains and 727 topics. Our evaluation protocol combines computational and rubric-based metrics with human judgment alignment. Results demonstrate the workflow's effectiveness in diverse setups, offering insights into LLM capabilities across domains. Our datasets and implementations are fully open-sourced.
format Preprint
id arxiv_https___arxiv_org_abs_2507_05528
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Conversational Education at Scale: A Multi-LLM Agent Workflow for Procedural Learning and Pedagogic Quality Assessment
Pei, Jiahuan
Ye, Fanghua
Sun, Xin
Deng, Wentao
Hindriks, Koen
Wang, Junxiao
Artificial Intelligence
Computation and Language
Large language models (LLMs) have advanced virtual educators and learners, bridging NLP with AI4Education. Existing work often lacks scalability and fails to leverage diverse, large-scale course content, with limited frameworks for assessing pedagogic quality. To this end, we propose WikiHowAgent, a multi-agent workflow leveraging LLMs to simulate interactive teaching-learning conversations. It integrates teacher and learner agents, an interaction manager, and an evaluator to facilitate procedural learning and assess pedagogic quality. We introduce a dataset of 114,296 teacher-learner conversations grounded in 14,287 tutorials across 17 domains and 727 topics. Our evaluation protocol combines computational and rubric-based metrics with human judgment alignment. Results demonstrate the workflow's effectiveness in diverse setups, offering insights into LLM capabilities across domains. Our datasets and implementations are fully open-sourced.
title Conversational Education at Scale: A Multi-LLM Agent Workflow for Procedural Learning and Pedagogic Quality Assessment
topic Artificial Intelligence
Computation and Language
url https://arxiv.org/abs/2507.05528