The knowledge graph
What do these labels mean?
- edge
- A typed relation between two concepts: one points at the other for a reason the model records.
- prerequisite_of
- One concept must be understood before the other. These decide study order.
- applies_to
- A rule applies to a concept. Administrative, not an order to study in.
- proposed
- A model suggested this edge. No person has confirmed it, so it is evidence of nothing yet.
- human-confirmed
- A person checked it against the source and accepted it.
- sequencing
- A prerequisite edge that affects the order lessons are offered in.
- structural
- A relation that describes the corpus rather than the order to read it.
The syllabus tree the learning paths are built from, and the prerequisite skeleton — the edges authored for those paths. Pick a node to read its content and the source it came from.
One graph, three confidence tiers. What is drawn here is the authored backbone — TP 15263 areas, their topics, and the hand-authored prerequisite skeleton. The regulation mesh has its own layer now: the mesh view, whose scope is what D9 measured — 77.4% of the deterministic extractor’s findings auto-admitted, recall 90.4%, precision 1.0 on a 40-edge sample. Proposed edges still stay in a reviewer view: the edge explorer is where a reader meets them, and it says how many it withholds. An edge between two objectives is drawn between the topics that own them, so several objectives can share one drawn edge.
The graph is the syllabus tree and the prerequisite skeleton — the edges authored for the learning paths. The LLM-assisted regulation mesh is not drawn here: proposed edges belong in a reviewer view.