Home▸AI & Machine Learning▸3D Maximal Marginal Relevance: Diverse Lesson Recommendations, Cosine Redundancy and the Relevance-Novelty Trade-off

📚 3D Maximal Marginal Relevance: Diverse Lesson Recommendations, Cosine Redundancy and the Relevance-Novelty Trade-off

A Blender-built library globe of lesson tiles on the unit sphere: place a query beacon, slide lambda from plain top-k to maximal marginal relevance and see near-duplicates drop out while topic coverage rises, checked against exhaustive search, rotation invariance and the lambda = 1 limit.

AI & Machine Learning3DModerate60 FPS📱 Mobile-adapted🟠 Blender⇄ 2D version
3D Maximal Marginal Relevance: Diverse Lesson Recommendations, Cosine Redundancy and the Relevance-Novelty Trade-off ↗ Open standalone
⚙ Under the hood

A Blender-built library globe of lesson tiles on the unit sphere: place a query beacon, slide lambda from plain top-k to maximal marginal relevance and see near-duplicates drop out while topic coverage rises, checked against exhaustive search, rotation invariance and the lambda = 1 limit.

Maximal Marginal RelevanceRecommender SystemsDiversityCosine SimilarityEmbeddingsInformation RetrievalEdTechMachine Learning

3D · Three.js / WebGL renderer · 60 FPS target · runs fully client-side, no install

What did you find?

Add reproduction steps (optional)