Knowledge Graph: Definition and Examples

A Knowledge Graph is a structured representation of information that models relationships between real-world entities. Unlike a classic database organized in tables, a Knowledge Graph uses triplets (subject → relation → object) to capture the richness of connections between concepts. For example, "Paris → is the capital of → France" or "GPT-4 → is developed by → OpenAI."

This technology is at the heart of modern search engines like Google, which has used its own Knowledge Graph since 2012 to enrich search results with contextual information. When you see an information box to the right of a Google search (birth date of a celebrity, population of a city), it is the Knowledge Graph providing that structured data.

In artificial intelligence, Knowledge Graphs play a crucial role in anchoring language models to verifiable facts. They help reduce hallucinations by offering a structured source of truth that the AI can consult. This hybrid approach — combining the generative power of LLMs with the factual precision of knowledge graphs — is one of the most promising avenues for improving the reliability of AI systems.

In the context of prompt engineering, understanding Knowledge Graphs allows you to better structure your queries by making relationships between concepts explicit. An informed user can ask an LLM to reason in graph terms, identify implicit connections between ideas, or build a structured representation of a knowledge domain from raw text.