Automated Domain Question Mapping (Dqm) With Educational Learning Materials

Jiho Noh, Kennesaw State University

Mukhesh Raghava Katragadda, Kennesaw State University

Dabae Lee, Kennesaw State University

Date

01/2026

AI Overview

Why it matters

Traditional concept maps lack concepts tailored for multi-level pedagogical goals, and labeled relationship data is scarce. Mapping questions instead of abstract concepts can better align instruction, assessment, and personalized learning systems.

Study at a glance

Question: Can language models generate educational questions and build structured question maps directly from textbooks?
Setting and data: 723 KhanQ and 86,821 answerable SQuAD question-answer pairs; one computer science textbook with 1,071 chunks and 7,275 pairs.
Method: Language model fine-tuning and evaluation, relationship classification, and graph post-processing with maximum spanning tree pruning.
AI tools: BART-large, Pegasus-large, T5-Large, GPT-2, GPT-o4-mini, Sentence-BERT

Key findings

Fine-tuned encoder-decoder models outperformed decoder-only models across question generation metrics, though GPT-o4-mini performed competitively in few-shot settings.
A fine-tuned GPT-2 Medium model achieved a 90% macro average F1-score when classifying hierarchical relationships between textbook text chunks.
Pruning combined question graphs using semantic similarity, relationship scores, and maximum spanning trees produced navigable, cycle-free educational question maps.

Summary drafted with AI from the full paper and may contain errors. See the full paper for details.

Concept maps have been widely utilized in education to depict knowledge structures and the interconnections between disciplinary concepts. Nonetheless, devising a computational method for automatically constructing a concept map from unstructured educational materials presents challenges due to the complexity and variability of educational content. We focus primarily on two challenges: (1) the lack of disciplinary concepts that are specifically designed for multi-level pedagogical purposes from low-order to high-order thinking, and (2) the limited availability of labeled data concerning disciplinary concepts and their interrelationships. To tackle these challenges, this research introduces an innovative approach for constructing Domain Question Maps (DQMs), rather than traditional concept maps. By formulating specific questions aligned with learning objectives, DQMs enhance knowledge representation and improve readiness for learner engagement. The findings indicate that the proposed method can effectively generate educational questions and discern hierarchical relationships among them, leading to structured question maps that facilitate personalized and adaptive learning in downstream applications.