A reproducible wine recommendation knowledge graph with logic- and embedding-based solvers, and an LLM grounding experiment that reduces hallucinated wineries 3.5x.
By chatting or signing in you agree to the Terms and chat-message logging (revocable in History).