Hybrid RAG

Hybrid RAG combines dense retrieval (embedding similarity) with sparse retrieval (BM25 keyword matching). This gives you the best of both worlds: semantic understanding AND exact keyword matching.

How it Works

1. Your query is processed two ways:
   • Dense: Converted to embedding vector for semantic search
   • Sparse: Keywords extracted for BM25 matching
2. Both result sets are merged using Reciprocal Rank Fusion (RRF)
3. Final results combine semantic relevance with keyword precision

RAG Wizard Configuration

Step 1: Project Setup

• Domain: Any domain (works well with mixed content)
• Use Case: "Document Analysis", "Q&A", or "Research"

Step 2: Data Sources

Hybrid RAG handles mixed-domain documents better than Naive RAG since BM25 catches exact matches.

Step 3: Document Processing

Step 4: Pipeline Configuration

BM25 Configuration

BM25 parameters can be tuned in the Pipeline tab:

Step 5: API Endpoints

Generate your API key and integrate using REST API, Python SDK, or MCP Protocol:

curl -X POST "https://api.guidedmind.ai/rag/search" \
-H "X-API-Key: rk_your_key_here" \
-H "Content-Type: application/json" \
-d '{
  "query": "ERROR-402 payment failed",
  "options": {
    "limit": 5,
    "threshold": 0.5,
    "search_method": "hybrid"
  }
}'

Performance Comparison

When to Use Hybrid RAG

When to Upgrade

Consider Graph RAG if:

• Your documents reference each other extensively
• Users ask multi-hop questions
• You need relationship discovery across documents