A network of entities (people, places, concepts, events) and the semantic relationships between them, used to represent and query structured knowledge.
Full definition
A knowledge graph is a structured representation of knowledge as a network of entities and the semantic relationships between them. Entities are discrete things — a person, a company, a concept, a location, a date — and relationships describe how those entities are connected. A knowledge graph encodes not just data but meaning: not just 'Alice' and 'project X', but 'Alice leads project X', enabling reasoning and inference that flat data cannot support.
The concept of knowledge representation as a graph dates to the semantic network theories of the 1960s (Ross Quillian's work on semantic memory) and the 1970s (Marvin Minsky's frames). The term 'knowledge graph' was popularised at scale by Google in 2012, when it launched the Google Knowledge Graph to power structured answers in search results. Enterprise knowledge graphs followed — IBM Watson, Microsoft's Cortana entity graph, LinkedIn's Economic Graph. The 2010s brought personal knowledge graphs to individuals through note-taking tools like Roam Research and Obsidian, which introduced bi-directional linking as a mechanism for personal graph construction. AI-native tools like Brinn now build personal knowledge graphs automatically from unstructured notes.
Entities are the nodes of a knowledge graph — the distinct things the graph knows about. In a personal knowledge graph, entities include people you know, projects you're working on, books you've read, locations you've visited, and recurring concepts in your thinking.
Relationships are the edges connecting entities — they describe how two entities relate to each other. 'Alice manages Project X', 'Project X depends on API Y', 'Book Z influenced Concept W' are all relationships. The richness of a knowledge graph comes from the diversity and depth of its relationships.
Modern knowledge graphs use vector embeddings to encode the semantic meaning of text, enabling similarity search rather than just keyword matching. This allows queries like 'what did I write about focus and deep work?' to surface relevant notes even if they don't contain those exact words.
Building a knowledge graph from unstructured text requires identifying and extracting entities automatically. Named Entity Recognition (NER) models identify people, organisations, locations, and dates. Brinn extends this with topic and concept extraction to build a richer personal graph.
How Brinn applies this
Brinn builds a personal knowledge graph automatically from your notes. Every time you capture a thought, Brinn's AI pipeline extracts entities, identifies relationships between them, and indexes the content semantically. Over time, your Brinn graph becomes a living map of your knowledge — people, projects, ideas, and the connections between them — without any manual linking.
Try it free