What is Knowledge Graph? Explained in Everyday Language for AI Beginners

Knowledge graphs are frequently mentioned in the context of building today’s AI systems. However, this concept originated in the 1980s. Before the era of LLMs, they were already quite common and increasingly integral to various applications and services across industries, such as search engines, social networks, digital assistants, and more. If you were to map out all the keywords and concepts on Wikipedia, you would get a huge knowledge graph too :)

An example of a knowledge graph reflecting a family reunion

So, first things first, what is a Knowledge graph? And how does it work in AI?

Imagine you’re trying to organize a big family reunion. To do this efficiently, you decide to create a giant chart. On this chart, you connect every family member to others using lines that describe their relationships — like who is married to whom, who are siblings, and who are cousins. This chart helps everyone understand how they’re connected within the family.

A Knowledge Graph works similarly but in the realm of data and information. Think of it as a massive chart where each point (or “node”) represents a piece of information, like a fact, concept, or object. The lines (or “edges”) between these points describe the relationships between them — how one piece of information is related to…