Behavior Prediction on GROVE Full Wikipedia domains 1.0
70.7Accuracy (Aero.)Human Profile
Evaluation Results
| Method | Links | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| Human ProfilePrediction Model=GPT-4.1, Evaluation Model=GPT-4.12026.04 | 70.7 | 65.4 | 63.1 | 68.1 | 55.2 | 62.5 | 55.6 | 53.9 | 63.9 | 62 | |
| CDT (Ours)Prediction Model=GPT-4.1, Evaluation Model=GPT-4.12026.04 | 70 | 69.1 | 68.5 | 70.8 | 74 | 65 | 66.7 | 64.5 | 70.1 | 68.7 | |
| Summarized-basedPrediction Model=GPT-4.1, Evaluation Model=GPT-4.12026.04 | 69.8 | 62.9 | 66.2 | 68.1 | 68.8 | 65 | 59.7 | 58.3 | 64.9 | 64.9 | |
| VanillaPrediction Model=GPT-4.1, Evaluation Model=GPT-4.12026.04 | 67.7 | 62.9 | 64.4 | 69.4 | 52.1 | 59.2 | 61.1 | 54.4 | 62.4 | 61.5 | |
| RAGPrediction Model=GPT-4.1, Evaluation Model=GPT-4.12026.04 | 67.3 | 62.5 | 64.4 | 66 | 60.4 | 59.2 | 55.6 | 55.2 | 67.5 | 62 |