Knowledge graphs: concepts, prerequisites and textbook evidence
Make concepts and relationships inspectable, and give questions, materials and learning records a traceable context.
Start with the evidence behind a relationship
In our question-bank research prototype, the model extracts concepts and relationships from a textbook section and retains source excerpts. Checks on concepts, links and cycles prepare the graph for human review.
A source excerpt supports inspection; it does not guarantee that a proposed semantic relationship is correct. Structural checks and human review both remain part of the workflow.
Concept structure and learner state are different information
Vispo’s prerequisite module also organizes concept relationships, removes duplicate links and cycles, and limits graph size and depth. This structure describes how learning content connects.
A learner-state module summarizes teacher-confirmed errors using time-weighted concept signals. This is an interpretable heuristic, not a calibrated probability of mastery; an absence of recorded errors does not establish learning.
Research foundations and evaluation
Hogan and colleagues’ Knowledge Graphs survey provides conceptual background on graph data models, structure and quality. Vispo combines textbook extraction with structural checks rather than reproducing one paper in full.
Our next research questions concern the quality of concept relationships, the evidence behind them and their usefulness when teachers choose the next exercise. Separately, KT4EQG’s work on knowledge tracing and exercise alignment asks whether an item truly requires its target knowledge components.