Knowledge-graph-for-Vietnam-education

Digital Transformation in Teaching and Learning: A Knowledge Graph as Foundational Infrastructure

A short proposal draft — 5 October 2026

1. The problem: digital materials without shared knowledge infrastructure

Education develops people: their knowledge, abilities, values, and capacity to act independently. This proposal focuses on teaching, learning, and assessment. Vietnam has already digitized substantial educational content, from textbooks and lesson slides to videos and examination papers. Yet making materials available online does not automatically make the knowledge inside them connected or reusable. The problem I propose to investigate is fragmentation: resources are organized around individual books, courses, and platforms, while their underlying concepts, skills, and dependencies often remain implicit.

The missing layer is a shared, structured knowledge database serving as infrastructure for educational applications. A teacher should be able to identify what a resource teaches; an assessment should make clear what knowledge and skills it probes; a learner should receive useful guidance about what to study next. A common knowledge structure could connect these activities. Establishing how much existing infrastructure already meets this need in Vietnam should be an early task, rather than assuming that nothing exists.

2. The proposed solution: an open knowledge graph for secondary education

I propose an open knowledge graph of upper-secondary school knowledge, beginning with Grade 12 mathematics. It would represent concepts with their definitions and notation, propositions with their assumptions and conclusions, and skills with the tasks that demonstrate them. Relationships would distinguish definitional dependence, dependence within a proof, and learning prerequisites. Lessons, examples, exercises, and assessment tasks would link to these objects. Every contribution would have an identifiable source, revision history, and review status.

A textbook selects, explains, and arranges knowledge into a purposeful learning pathway. The graph would allow different textbooks and pathways to reference a shared underlying structure while preserving alternative explanations and approaches. Teachers and students would have a place to use this resource and contribute examples, questions, solutions, corrections, and feedback. Public access would be accompanied by appropriate licenses and quality review.

3. Why this should be an early foundational step

The proposal addresses two bottlenecks. First, common identifiers and explicit relationships would provide a map for connecting large collections of teaching materials and examination questions. Resources could be found by the knowledge they address, not just their filenames or chapter labels. Second, mapping assessment tasks to concepts and skills would provide a basis for interpreting evidence about a learner. If a student struggles with an extremum problem, targeted questions could help distinguish a gap in differentiation, misunderstanding of an extremum condition, or difficulty with algebra. The graph supports that diagnosis; response data and an assessment model are still needed to make it.

There are useful precedents. Vietnam’s project sequencing more than 1,000 Vietnamese genomes and its population and identity databases illustrate investment in foundational data, although their objectives and governance differ from education.[1][2] Internationally, OntoMathEdu models school mathematics and prerequisites, EDUKG provides a public K–12 knowledge resource, and the CASE standard enables shared references to learning standards and competencies.[3][4][5] These precedents support feasibility. They do not prove that a graph alone improves learning. The argument for starting early is architectural: agreeing on a small, usable common structure allows later resources and applications to accumulate coherently, without waiting for a complete national graph.

4. What could change if it works

At system level, this infrastructure could support connected resource discovery, question-bank organization, and more transparent assessment coverage. It could contribute to improving national upper-secondary assessments by making the knowledge and skills tested easier to inspect. Question difficulty, however, must be calibrated using learner responses; validity, reliability, and fairness require their own evaluation. Counting graph nodes or attaching labels to questions would not establish assessment quality.

At learner level, the graph could be combined with evidence to estimate a changing knowledge state: what is mastered, uncertain, or ready to learn. That would support targeted assistance and reassessment. ALEKS demonstrates knowledge-state assessment, while ASSISTments provides experimental evidence for a package of feedback and teacher support.[6][7] Neither establishes the effect of this proposed open graph. Following learning from childhood into adulthood is a longer-term aspiration requiring broader content coverage, updated evidence, and protection of personal records. The knowledge resource can be open while individual learning records remain protected.

5. A practical one-to-five-year pathway

The first year should produce a small, reviewed Grade 12 mathematics pilot: a documented schema, stable identifiers, sample content, linked questions, and a simple interface for use and contribution. Teachers and students should test whether it helps them find relevant resources and identify learning difficulties. The next stage, over years two and three, should expand coverage, refine the structure, establish contributor recognition and review procedures, and evaluate classroom applications with partner schools and researchers.

Over years four and five, expansion across grades and subjects should depend on demonstrated usefulness, sustainable funding, and effective governance. Success should be measured through data quality, reuse, teacher workload, diagnostic accuracy, and learning outcomes. The immediate invitation is concrete: build a small resource together, test it, and improve it. My central claim is that structured knowledge data deserves to be an early infrastructure priority in the digital transformation of teaching and learning. A shared knowledge graph is the proposed starting point for turning scattered digital materials into a resource that grows through use and supports learner progress.

References

1. Government of Vietnam, Completion of the project sequencing 1,000 Vietnamese genomes.

2. Government of Vietnam, Building the national population database.

3. Falileeva et al. (2020), OntoMathEdu: Prerequisites, Educational Levels and Educational Projections.

4. Tsinghua University Knowledge Engineering Group, EDUKG repository.

5. 1EdTech, Introduction to the CASE Standard.

6. Doignon and Falmagne (2015), Knowledge Spaces and Learning Spaces.

7. SRI International, ASSISTments Efficacy and Replication Studies.