Instructor-curated knowledge graph and dual-readability typed chunk corpus for a dimensional modelling course¶

Design · All designs
Evidence · no claims cited

Description¶

The course knowledge is represented as a lightweight knowledge graph containing 43 concepts organised into thematic groups, connected through structural relations (is a, part of) and didactic relations (precedes, confused with, mutually exclusive with) curated by the instructor. The course material is authored as structured Markdown revised "for dual readability: human and machine", with each concept on a dedicated page, content segmented into nine chunk types determined by the markup itself, and every chunk indexed in a ChromaDB store with concept id and chunk type as filterable attributes. Every answer is grounded in this instructor-authored material with permalinks to the relevant teaching pages.

Design Implications¶

Context¶

Requirements¶

Each new content source must be indexed with concept identifiers and chunk types so constrained retrieval continues to operate on pedagogical role

Constraints¶

The current corpus is mono-authored, originating solely from the instructor's teaching notes

Target Learners¶

University students in the dimensional modelling course

Learning Goals¶

Grounding chatbot answers in verified curricular content with permalinks to teaching pages

Claims¶

Deterministic Pedagogical Orchestration Architecture [+M]

Chain-of-Concept: topic-graph organisation of past assessments replacing similarity-based retrieval

Examples¶

-

Key Sources¶

Laurent Brisson, Maria Teresa Segarra and Gregory Smits. (2026). A didactical-driven teacher assistant for a dimensional modeling course. https://arxiv.org/abs/2607.22598

Page metadata

Field	Value
`type`	`design`
`id`	`instructor-kg-typed-chunk-corpus`
`title`	`Instructor-curated knowledge graph and dual-readability typed chunk corpus for a dimensional modelling course`
`description`	`The course knowledge is represented as a lightweight knowledge graph containing 43 concepts organised into thematic groups, connected through structural relations (is a, part of) and didactic relations (precedes, conf...`
`status`	`review`
`generated`	by `process:settle-candidates` on `2026-10-10`
`trust`	unverified — no one has confirmed this page's content
`sources`	1 — `laurent-brisson-2026`

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