A Framework for Automated Construction of Prerequisite-Structured Domain Ontologies from Textbooks
MSc Thesis Defense by: Faizaan Naeem Lakdawala
Date: 21st September 2026
Time: 12 PM
Location: Essex Hall 122
Abstract:
Intelligent Tutoring Systems require domain ontologies that encode both structured learning content and prerequisite relationships between topics to guide learners through a domain. However, existing automated ontology construction approaches produce only semantic and domain-factual relationships, not the pedagogically meaningful prerequisite ordering that tutoring systems need. Meanwhile, automated prerequisite prediction methods exist as a separate research area, with their outputs never integrated into ontology construction. Implementations that combine prerequisite structures with ontologies rely entirely on manual construction by domain experts, making them unable to scale across subjects.
This thesis proposes a 7-stage automated framework that takes a textbook and a domain dataset as inputs and produces a complete, prerequisite-structured domain ontology ready for deployment in an Intelligent Tutoring System. The framework integrates topic extraction, automated prerequisite prediction, LLM-assisted prerequisite refinement, and meta-ontology-guided ontology construction into a single cohesive workflow, replacing what previously required extensive manual expert effort. An expert-in-the-loop review mechanism balances automation with human quality assurance. The resulting ontologies can be directly applied in personalized e-learning systems to support scalable and adaptive education.
Keywords: Ontology Construction, Prerequisite Prediction, Intelligent Tutoring Systems, Knowledge Space Theory, Meta-Ontology
Thesis Committee:
Internal Reader: Dr. Olena Syrotkina
Internal Reader: Dr. Jessica Chen
Advisor: Dr. Xiaobu Yuan
Chair: Dr. Andreas Maniatis