Applications and use cases of graph RAG in LLMs
Graph-based RAG can be particularly effective in various applications:
- Question-answering over knowledge graphs
- Personalized recommendation systems
- Scientific literature analysis
- Drug discovery and biomedical research
- Social network analysis
Here’s an example of how graph RAG could be used for a recommendation system:
class RecommendationGraphRAG(GenerativeGraphRAG):
def get_recommendations(
self, user_id: str, num_recommendations: int = 5
) -> List[str]:
user_node = self.retrieve(f"User {user_id}", k=1)[0]
user_interests = self.graph.nodes[user_node].
get('interests', [])
...
