graph TD
A[User Query] --> B[RAG System]
B --> C[Query Embedding]
C --> D[Vector Database]
D --> E[Retrieve Similar Vectors]
E --> F[Fetch Corresponding Documents]
F --> G[Augment Original Query]
G --> H[LLM]
H --> I[Generate Response]
I --> J[Return Answer to User]
subgraph "Document Ingestion"
K[Source Documents] --> L[Text Chunking]
L --> M[Embedding Generation]
M --> N[Store in Vector Database]
end
D -.-> N
4 Database models
Wikipedia (2016)
4.1 Document databases
Most well known example is mongodb. Document databases mostly store json documents. They are schema-free organization of data. That is unlike the relational databases, you do not upfront have to design database schema.
This has advantages and disadvantages.
Examples include
- MongoDB
- Databricks
- Amazon DynamoDB
- Microsoft Azure Cosmos DB
- Couchbase
- Firebase (google)
- Oracle NoSQL
4.2 Graph Database
