Personalized User Simulation on Twitter
0.7864Sim FidelityTWICE
Evaluation Results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| TWICEBackbone=ChatGPT-5.5, Baseline strategy=event2025.12 | 0.7864 | 20.22 | 3.487 | 0.2319 | 21.93 | |
| TWICEBackbone=ChatGPT-5.5, Baseline strategy=general2025.12 | 0.7789 | 20.12 | 3.22 | 0.2248 | 19.8 | |
| Memory-onlyBackbone=ChatGPT-5.52025.12 | 0.7713 | 20.4 | 3.235 | 0.1942 | 17.76 | |
| OASISBackbone=ChatGPT-5.52025.12 | 0.7586 | 20.53 | 3.436 | 0.213 | 20.43 | |
| PopAlign personaBackbone=ChatGPT-5.52025.12 | 0.749 | 21.01 | 3.306 | 0.2718 | 26.9 | |
| Implicit profileBackbone=ChatGPT-5.52025.12 | 0.7471 | 21.51 | 3.464 | 0.2286 | 18.88 | |
| Profile-onlyBackbone=ChatGPT-5.52025.12 | 0.7373 | 21.11 | 3.832 | 0.2154 | 27.62 | |
| TWICEBackbone=Qwen3-14B, Baseline strategy=event2025.12 | 0.7275 | 1.102 | 0.7479 | 0.2959 | 12.93 | |
| Implicit profileBackbone=Qwen3-14B2025.12 | 0.7247 | 3.198 | 0.4079 | 0.2562 | 8.9 | |
| TWICEBackbone=Qwen3-8B, Baseline strategy=event2025.12 | 0.7223 | 4.822 | 0.9456 | 0.291 | 17.93 | |
| TWICEBackbone=Qwen3-8B, Baseline strategy=general2025.12 | 0.7215 | 5.746 | 1.11 | 0.3021 | 19.58 | |
| Implicit profileBackbone=Qwen3-8B2025.12 | 0.7214 | 0.402 | 0.6424 | 0.2443 | 3.34 | |
| TWICEBackbone=Qwen3-14B, Baseline strategy=general2025.12 | 0.7205 | 1.412 | 0.8443 | 0.2914 | 13.57 | |
| Memory-onlyBackbone=Qwen3-14B2025.12 | 0.7201 | 3.21 | 0.2566 | 0.2772 | -2.98 | |
| Non-personalizedBackbone=Qwen3-14B2025.12 | 0.7155 | 6.708 | 0.2941 | 0.2997 | -0.33 | |
| Memory-onlyBackbone=Qwen3-8B2025.12 | 0.7154 | 4.299 | 0.0292 | 0.2663 | -4.53 | |
| OASISBackbone=Qwen3-14B2025.12 | 0.7152 | 8.345 | 0.9226 | 0.2967 | 2.5 | |
| OASISBackbone=Qwen3-8B2025.12 | 0.7102 | 5.366 | 0.4148 | 0.3313 | 0.92 | |
| Profile-onlyBackbone=Qwen3-14B2025.12 | 0.7097 | 9.204 | 0.9563 | 0.2918 | 6.94 | |
| PopAlign personaBackbone=Qwen3-14B2025.12 | 0.6969 | 12.41 | 1.3 | 0.4137 | 11.51 | |
| PopAlign personaBackbone=Qwen3-8B2025.12 | 0.6931 | 9.903 | 0.9826 | 0.4442 | 11.24 |