Knowledge Base
Definition
A knowledge base is an organised collection of information designed to support storage, retrieval and use of knowledge by people or software systems. It may contain documents, records, facts, structured data or other information sources.
Why it matters
Knowledge bases are common external information sources for search, RAG and agentic systems. They provide a managed corpus that can be queried at runtime instead of relying only on information encoded in model parameters.
How it works
Information storage
- A knowledge base can contain structured or unstructured information.
- Content may be organised through metadata, schemas, categories or other indexing structures.
Retrieval access
- Search or retrieval systems query the knowledge base to locate relevant information.
- The retrieval method can be lexical, semantic, vector-based, structured or hybrid.
Runtime use
- Retrieved knowledge can be supplied to an LLM, agent or application.
- Updates to the knowledge base can make current information available without retraining the model.
How Netsleek uses the term
Netsleek uses Knowledge Base as the broader information-source concept and keeps it distinct from Knowledge Graph. A knowledge base may contain documents or records without explicitly representing the relationships between entities as a graph.
Comparisons
- Knowledge Base vs Knowledge Graph: A knowledge base stores knowledge broadly. A knowledge graph explicitly models entities and relationships as a graph structure.
- Knowledge Base vs Model Knowledge: A knowledge base is an external runtime source. Model knowledge is encoded in learned parameters.
- Knowledge Base vs Database: A database stores data. A knowledge base is organised specifically to support knowledge access and use, and may span multiple data forms.
How Knowledge Base Connects Across AI Systems
Knowledge Base does not operate in isolation. It connects to other established concepts across the AI search and retrieval process:
- Supports retrieval: Retrievers can search or query a knowledge base for relevant information.
- Supports RAG: Knowledge bases are common evidence sources for Retrieval-Augmented Generation.
- Can include entities: Entity Linking can connect text mentions to canonical entities stored in a knowledge source.
- Can complement knowledge graphs: A knowledge base can contain or reference a Knowledge Graph.
Related glossary concepts
- Knowledge Graph
- Retrieval-Augmented Generation (RAG)
- Semantic Retrieval
- Vector Search
- Entity Linking
- Structured Data
- Machine-Readable Structure
Common misinterpretations
- A knowledge base is not automatically a knowledge graph.
- A knowledge base is not the same as an LLM’s learned parameters.
- Storing information in a knowledge base does not guarantee that a retriever will surface it.
Summary
A knowledge base is an organised external source of information that can be searched or queried by AI systems. It supports retrieval and grounding without requiring the underlying model to be retrained whenever the information changes.