Knowledge Graph
Authoritative Definition
A structured, explicit network representation of real-world entities, concepts, and relationships stored as nodes and edges in a graph database.
Overview & Technical Description
Knowledge Graphs encode domain-specific knowledge deterministically, organizing information into interconnected nodes (entities) and edges (relationships). By providing a formal semantic structure, they enable complex, multi-hop reasoning, strict factual grounding, and formal querying using languages like SPARQL or Cypher. In the context of LLMs and AI agents, Knowledge Graphs act as an authoritative source of truth. They provide a vital counterbalance to the probabilistic nature of vector embeddings by offering explicit, non-lossy semantic relationships. This structural context helps guide an agent's reasoning process, allowing it to traverse explicit paths to discover non-obvious connections between disparate entities. Integration with generative AI often involves converting unstructured text into triples (subject, predicate, object) to populate the graph, or utilizing the graph to dynamically construct highly contextual prompts. This synergy powers advanced architectures like GraphRAG, enhancing explainability and significantly reducing hallucinations.
Editorial Notes
Provides explicit semantic relationships that vector embeddings cannot represent. Building high-quality knowledge graphs from unstructured data remains a significant data engineering challenge, often requiring robust entity resolution and ontology management pipelines.