AI Society Corpus
An interactive knowledge graph for NUS AI Society that links AI/ML study topics by prerequisites, shared tags, and backlinks.
- Status
- active · NUS AI Society
- Role
- Initial author and main contributor: I wrote the first build and 30 of the 82 commits in the repository as of August 2026; other society members contribute topics and fixes.
- Stack
- knowledge-graph, astro, react, d3, ai-education
- Repository
- Source code
- Dates
Problem
AI/ML learners need to see how concepts depend on and relate to one another, not just read a long list of topics.
What I built
Topics live in validated Markdown with structured frontmatter. A static Astro build derives prerequisite, tag-similarity (Jaccard), and WikiLink-backlink edges and renders them with React and a D3 force graph.
Evidence
- Live site: aisoc-corpus.aisocietysoc.workers.dev
- Content validation, unit tests, and Playwright checks run in the repository
- 6 topic files published at the time of writing
One hard decision
Relationships are computed at build time from the Markdown rather than curated by hand, so a new topic joins the graph through its frontmatter alone.
Where it stands
Readers can move through a navigable concept network, and contributors get a reviewable workflow with validation.
Limits
The corpus is small so far, and tag-similarity edges are inferred: two topics that share tags are linked even when the relationship is weak.