Research map / Reinforcement learning / Visual and representation RL
KuaiSim: A Comprehensive Simulator for Recommender Systems
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Use KuaiSim instead of competing simulators to train DDPG | DDPG-trained recommender · KuaiRand simulator comparison | Total reward: 9.570 $\pm$ 0.077 -> 10.081* $\pm$ 0.116 | Helped |
| Use KuaiSim instead of competing simulators to train DDPG | DDPG-trained recommender · KuaiRand simulator comparison | AUC: 0.6929 $\pm$ 0.0019 -> 0.7234* $\pm$ 0.0021 | Helped |
| Use HAC rather than other tested algorithms | Whole-session recommendation · KuaiRand | Total reward: 10.0873 $\pm$ 0.0571 -> 10.1742 $\pm$ 0.0634 | Helped |
| Use ListCVAE rather than CF or PRM | Request-level recommendation · KuaiRand | Coverage: 446.100 $\pm$ 15.648 vs. 100.969 $\pm$ 7.193 and 27.520 $\pm$ 3.210 | Helped |
| Use RLUR rather than TD3 or CEM | Cross-session retention optimization · KuaiRand | User retention: 0.581 $\pm$ 0.001 -> 0.607 $\pm$ 0.002 | Helped |
| Use HAC rather than other tested algorithms | Whole-session recommendation · ML-1m | Total reward: 7.5333 $\pm$ 0.0273 -> 7.5791 $\pm$ 0.0206 | Helped |
| Set max time step to 20 rather than other tested settings (ablation) | TD3 with KuaiSim · KuaiRand | Model performance: 20 | Helped |
| Set slate size to 20 rather than other tested settings (ablation) | TD3 with KuaiSim · KuaiRand | Model performance: 20 | Helped |
Closest papers by meaning
- LLM-Powered User Simulator for Recommender System (AAAI 2025 · similarity 0.91)
- Think before Recommendation: Autonomous Reasoning-enhanced Recommender (NeurIPS 2025 · similarity 0.89)
- On the Unexpected Effectiveness of Reinforcement Learning for Sequential Recommendation (ICML 2024 · similarity 0.89)
- Would You Like Your Data to Be Trained? A User Controllable Recommendation Framework (AAAI 2024 · similarity 0.88)
- ConvRec-R1: Training LLM-based Conversational Recommender Systems with Reinforcement Learning (ICLR 2026 · similarity 0.88)
- R$^2$ec: Towards Large Recommender Models with Reasoning (NeurIPS 2025 · similarity 0.88)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.