Skip to content

Knowledge Graph (Neo4j)

RagLeap uses Neo4j for knowledge graph storage, enabling more accurate and contextual answers than traditional vector search alone.

What is a Knowledge Graph?

A knowledge graph maps relationships between concepts in your documents. Instead of just finding similar text, RagLeap understands how concepts relate to each other.

Example: - Traditional search: finds "product pricing" text - Knowledge graph: understands that Product A → costs ₹999 → includes Feature X → requires Setup Y

How It Works

When you upload a document, RagLeap: 1. Extracts entities (people, products, dates, amounts) 2. Maps relationships between entities 3. Stores in Neo4j graph database 4. Combines with vector search for hybrid retrieval

Benefits

  • More accurate answers to complex questions
  • Better handling of multi-hop queries ("what products does our enterprise plan include?")
  • Entity linking across multiple documents
  • 94.3% retrieval accuracy (vs ~78% vector-only)

Configuration

Neo4j is included on all plans including free tier. No configuration needed — it works automatically.

Self-Hosted

On the self-hosted plan, Neo4j runs on your own server. Your knowledge graph never leaves your infrastructure.