Vector Search with Graph Context
Vector Embedding plays a crucial role in enhancing search capabilities:
Comparison of Vector-Only vs. Graph-RAG:
- Vector-only searches may lack context, while Graph-RAG utilizes graph traversal to provide richer, multi-step context.
- This leads to more complex and informative responses.
Contextual Prompts:
- Context is used to answer prompts (in JSON format). With graph traversal, this context involves more steps, allowing for more elaborate retrieval queries.
Node Embedding
Useful in GraphRAG is understanding the relationships of nodes in a Knowledge Graph using node embeddings.
