Retrieval-Augmented Generation (RAG)

RAG (Retrieval-Augmented Generation) search allows you to retrieve relevant chunks from your collections based on a query. This enables language models to generate responses grounded in your specific documents and knowledge base.

Search Methods

OpenGateLLM supports multiple search methods:

Method	Description
semantic	Vector similarity search using embeddings
lexical	Keyword-based search (BM25)
hybrid	Combination of semantic and lexical search

Search Parameters

• prompt: Search query (required)
• collections: List of collection IDs to search in (required)
• method: Search method (default: semantic)
• limit: Number of results to return (default: 10, max: 200)
• offset: Pagination offset (default: 0)
• rff_k: RRF constant for hybrid search (default: 20)
• score_threshold: Minimum similarity score (0.0-1.0, only for semantic)
• web_search: Add internet search results (default: false)
• web_search_k: Number of web results (default: 5)

Search Flow

Performing Searches

Semantic search

curl -X POST http://localhost:8000/v1/search \
  -H "Authorization: Bearer <api_key>" \
  -H "Content-Type: application/json" \
  -d '{
    "prompt": "What is machine learning?",
    "collections": [1, 2],
    "method": "semantic",
    "limit": 10,
    "score_threshold": 0.7
  }'

Hybrid search

curl -X POST http://localhost:8000/v1/search \
  -H "Authorization: Bearer <api_key>" \
  -H "Content-Type: application/json" \
  -d '{
    "prompt": "Python programming",
    "collections": [1],
    "method": "hybrid",
    "limit": 10,
    "rff_k": 20
  }'

With web search

curl -X POST http://localhost:8000/v1/search \
  -H "Authorization: Bearer <api_key>" \
  -H "Content-Type: application/json" \
  -d '{
    "prompt": "Latest AI developments",
    "collections": [1],
    "method": "semantic",
    "limit": 10,
    "web_search": true,
    "web_search_k": 5
  }'

info

See Configuration for more details.

Web Search Integration

When web_search is enabled, OpenGateLLM:

1. Generates a web search query from your prompt
2. Retrieves results from the configured web search engine
3. Creates a temporary collection to store web results
4. Parses and processes each web result as a document
5. Performs the search across both your collections and web results
6. Automatically deletes the temporary web collection after returning results

info

Web search integration requires a web search engine to be configured. See Configuration for more details.

Next Steps

• Learn how to create and manage collections: Collections
• Learn how to import and process documents: Parsing and Chunking