
| Framework | Details |
|---|---|
| Method | Explore-then-Determine (EtD) |
| Components | Large Language Models (LLMs) + Graph Neural Networks (GNNs) |
| Task | Question Answering over Knowledge Graphs |
| Results | State-of-the-art performance with faithful reasoning |
The EtD framework synergizes LLMs with GNNs for efficient reasoning over complex Knowledge Graphs, enhancing precision and reducing resource costs. By exploring relevant information through GNNs and guiding LLMs in answering questions, EtD achieves high accuracy in KG Question Answering tasks. This framework showcases the potential of combining LLMs with specialized models for improved reasoning in complex knowledge domains.