Retrieval & RAG
Retrieval & RAG
API endpoints for document retrieval and RAG operations
Retrieval and RAG API
The Retrieval and RAG API allows you to query your knowledge base, retrieve relevant document chunks, and generate answers using Retrieval Augmented Generation (RAG).
Query Knowledge Base (POST)
Query the knowledge base and retrieve relevant chunks.
Request Body
| Parameter | Type | Description | Default |
|---|---|---|---|
| query | string | The search query | Required |
| vector_db | string | Vector database to use | ”pinecone” |
| top_k | integer | Number of results to return | 3 |
| model | string | Model to use for generating answers | ”gpt-4” |
Response
Returns relevant document chunks and potentially a generated answer.
Query Knowledge Base (GET)
Get relevant chunks from the knowledge base without generating an answer.
Query Parameters
| Parameter | Type | Description | Default |
|---|---|---|---|
| query | string | The search query | Required |
| vector_db | string | Vector database to use | ”pinecone” |
| top_k | integer | Number of results to return | 3 |
Response
Returns relevant document chunks without generating an answer.
RAG Query
Handle RAG query requests compatible with the RAG pipeline frontend.
Request Body
| Parameter | Type | Description | Default |
|---|---|---|---|
| query | string | The search query | Required |
| vector_db | string | Vector database to use | ”pinecone” |
| top_k | integer | Number of results to return | 3 |
| model | string | Model to use for generating answers | ”gpt-4” |
Response
Returns a structured response with:
- query: The original query
- answer: The generated answer
- chunks: Array of retrieved chunks with:
- id: Chunk ID
- text: Chunk text
- score: Relevance score
- source: Source document
- entities: Array of entity mentions (optional)
- total_chunks_searched: Number of chunks searched
- retrieval_time_ms: Time taken for retrieval in milliseconds
Simple Chatbot Query
Simple query endpoint for chatbot functionality.
Query Parameters
| Parameter | Type | Description |
|---|---|---|
| query_str | string | The search query |
| rag | boolean | Whether to use RAG |
| topk | integer | Number of results to return |
Response
Returns a response from the chatbot, optionally using RAG if specified.