Skip to content

GPT-4-extracted curriculum prerequisite knowledge graph (54 topic nodes, 47 directed prerequisite edges)

Design · All designs
Evidence · no claims cited

Description

A prerequisite knowledge graph extracted by GPT-4 from 26 lesson PDFs of a graduate-level AI course, following a three-stage protocol of topic extraction, boundary consolidation, and edge extraction. The graph contains "54 topic nodes and 47 directed prerequisite edges" and serves two roles: its topic nodes provide the closed label space for classification and survey alignment, and its directed prerequisite edges encode curriculum structure for future per-student gap propagation. The authors describe it as "a low-cost, instructor-auditable way to obtain a topic–edge artifact".

Design Implications

Context

Requirements

  • A course instructor with domain expertise to run and audit the GPT-4 extraction protocol over lesson PDFs

Constraints

  • The authors do not claim the graph is the unique correct structure; only unambiguous, non-bidirectional GPT-4 prerequisite judgments are retained

Target Learners

  • graduate-level AI course students

Learning Goals

  • mapping student questions to curriculum topics
  • identifying topic-level knowledge gaps

Claims

Examples

-

Key Sources

  • Youssef Medhat, Junsoo Park, Ploy Thajchayapong and Ashok K. Goel. (2026). Detecting Knowledge Gaps from Conversational AI Interactions Using Curriculum Prerequisite Graphs. CSEDM'26: 10th Workshop on Computer Science Educational Data Mining. https://dilab.gatech.edu
Page metadata
Field Value
type design
id gpt4-curriculum-prerequisite-knowledge-graph
title GPT-4-extracted curriculum prerequisite knowledge graph (54 topic nodes, 47 directed prerequisite edges)
description A prerequisite knowledge graph extracted by GPT-4 from 26 lesson PDFs of a graduate-level AI course, following a three-stage protocol of topic extraction, boundary consolidation, and edge extraction.
status review
generated by process:settle-candidates on 2026-10-11
trust unverified — no one has confirmed this page's content
sources 1 — youssef-medhat-2026

Frontmatter is the page's machine-readable record — see the Schema & Guide.