Knowledge Graph Question Answering on QALD-10en Wikidata-based (test)
74.4Hits@1SoG
Evaluation Results
| Method | Links | |
|---|---|---|
| SoGBackbone=GPT-4o2025.10 | 74.4 | |
| SpinachBackbone=GPT-4o2025.10 | 63.1 | |
| ToGBackbone Model=Qwen 2.5-72B2025.09 | 59.66 | |
| ToGBackbone Model=Qwen 3-235B2025.09 | 57.76 | |
| Think-on-GraphBackbone=GPT-42025.10 | 54.7 | |
| PoGBackbone Model=Qwen 3-235B2025.09 | 52.55 | |
| EffiQABackbone=GPT-42025.10 | 51.4 | |
| ToGBackbone Model=Qwen 2.5-7B2025.09 | 50.95 | |
| PoGBackbone Model=Qwen 2.5-72B2025.09 | 48.18 | |
| Self-ConsistencyBackbone=GPT-4o2025.10 | 47.2 | |
| PoGBackbone Model=Qwen 2.5-7B2025.09 | 46.07 | |
| IO PromptBackbone=GPT-4o2025.10 | 45.3 | |
| Think-on-GraphBackbone=GPT-4o, Reproduced status=reproduced2025.10 | 45.2 | |
| RoGBackbone Model=Qwen 2.5-72B2025.09 | 44.94 | |
| Chain-of-ThoughtBackbone=GPT-4o2025.10 | 44.8 | |
| KG-R1Backbone Model=Qwen2.5 3B, Training Dataset=WebQSP2025.09 | 41.96 | |
| ReknoSBackbone Model=Qwen 2.5-72B2025.09 | 39.94 | |
| ReknoSBackbone Model=Qwen 3-235B2025.09 | 39.44 | |
| ReknoSBackbone Model=Qwen 2.5-7B2025.09 | 37.94 | |
| RoGBackbone Model=Qwen 3-235B2025.09 | 37.54 | |
| KG-R1Backbone Model=Qwen2.5 3B, Training Dataset=CWQ2025.09 | 34.74 | |
| SCBackbone Model=Qwen 3-235B2025.09 | 29.53 | |
| RoGBackbone Model=Qwen 2.5-7B2025.09 | 29.33 | |
| CoTBackbone Model=Qwen 3-235B2025.09 | 28.43 | |
| IOBackbone Model=Qwen 3-235B2025.09 | 24.12 | |
| SCBackbone Model=Qwen 2.5-72B2025.09 | 21.62 | |
| CoTBackbone Model=Qwen 2.5-7B2025.09 | 21.05 | |
| CoTBackbone Model=Qwen 2.5-72B2025.09 | 20.52 | |
| SCBackbone Model=Qwen 2.5-7B2025.09 | 20.32 | |
| IOBackbone Model=Qwen 2.5-72B2025.09 | 17.45 | |
| IOBackbone Model=Qwen 2.5-7B2025.09 | 17.43 |