Lesson 304
Knowledge Graphs & the Semantic Web
RDF · ontologies · SPARQL · linked data
1:00How facts are stored as subject–predicate–object triples, linked by URIs into a queryable, reasoning-capable graph — and how embeddings predict the links that are missing.
By the end, you can
- Identify the subject, predicate, and object in any triple and explain how multiple triples form a graph.
- Read Turtle notation and explain what a namespace prefix does.
- Distinguish TBox (ontology/schema) from ABox (instance data) and explain why both are stored as triples.
- Trace a simple inference step (subclass transitivity) and explain why the derived fact was never written down.
- Write and interpret a basic SPARQL SELECT query with triple patterns and variables.
- Explain the TransE equation h + r ≈ t and describe how it supports link prediction.
- Name three real-world knowledge graphs and their key characteristics.
- Explain how grounding an LLM on a knowledge graph reduces hallucination and enables multi-hop reasoning.
Up next in AI, Machine Learning & Course Review
