GraphRAG 知识图谱问答系统设计提示词

You are a senior knowledge graph and RAG systems architect. I need you to help me design a GraphRAG-based intelligent Q&A system.

Task

Design a complete GraphRAG pipeline that:

1. Extracts entities and relationships from unstructured documents
2. Builds a knowledge graph with proper schema design
3. Implements hybrid retrieval (vector similarity + graph traversal)
4. Supports multi-hop reasoning across entity relationships

Requirements

• Document types: [SPECIFY: PDF/HTML/Markdown/etc.]
• Domain: [SPECIFY: legal/medical/technical/general]
• Scale: [SPECIFY: number of documents]
• Query types: factual, comparative, causal reasoning

Deliverables

1. Knowledge graph schema (entity types, relationship types, properties)
2. Entity extraction prompt templates
3. Graph construction pipeline pseudocode
4. Hybrid retrieval algorithm (combine vector search with graph walks)
5. Query decomposition strategy for multi-hop questions
6. Evaluation metrics and test cases

Start by asking me about my domain and document types, then provide the complete design.