Research map / Label-efficient and robust learning / Recognition and long-tail
Can Recommender Systems Teach Themselves? A Recursive Self-Improving Framework with Fidelity Control
Luankang Zhang, Hao Wang, Zhongzhou Liu, Mingjia Yin, Yonghao Huang, Jiaqi Li, Wei Guo, Yong Liu, Huifeng Guo, Defu Lian, Enhong Chen
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (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 |
|---|---|---|---|
| add fidelity-based quality control (ablation) | RSIR (SASRec) · amazon-sport, iteration 1 | NDCG@10: 0.0273 -> 0.0293 | Improved |
| add fidelity-based quality control (ablation) | RSIR (SASRec) · amazon-sport, iteration 2 | NDCG@10: 0.0209 -> 0.0294 | Improved |
| add fidelity-based quality control (ablation) | RSIR (SASRec) · amazon-sport, iteration 3 | Recall@10: 0.0210 -> 0.0528 | Improved |
| add fidelity-based quality control (ablation) | RSIR (SASRec) · amazon-sport, iteration 1 | Recall@10: 0.0472 -> 0.0512 | Improved |
| apply one RSIR iteration to SASRec | SASRec · amazon-beauty | NDCG@10: 0.0290 -> 0.0322 | Improved |
| apply RSIR to HSTU | HSTU · amazon-toys | Recall@10: 0.0869 -> 0.0924 | Improved |
| add RSIR instead of DR4SR augmentation | SASRec · amazon-sport | Recall@10: 0.0495 -> 0.0512 | Improved |
| run additional RSIR iterations (1 to 3) (ablation) | RSIR (SASRec) · amazon-sport | Recall@10: 0.0512 -> 0.0528 | Improved |
| run RSIR for 8 iterations vs 3 (ablation) | RSIR (SASRec) · amazon-sport | Recall@10: 0.0528 -> 0.0513 | Hurt |
| fidelity threshold tau=1 (too strict) (ablation) | RSIR · amazon-sport | Recall@10: 0.0512 -> 0.0469 | Hurt |
| exploitation probability p=1 vs p=0.5 (ablation) | RSIR · amazon-sport | Recall@10: 0.0512 -> 0.0478 | Hurt |
| exploitation probability p=0 vs p=0.5 (ablation) | RSIR · amazon-sport | NDCG@10: 0.0293 -> 0.0281 | Hurt |
Closest papers by meaning
- LLM4RSR: Large Language Models as Data Correctors for Robust Sequential Recommendation (AAAI 2025 · similarity 0.88)
- Would You Like Your Data to Be Trained? A User Controllable Recommendation Framework (AAAI 2024 · similarity 0.87)
- Think before Recommendation: Autonomous Reasoning-enhanced Recommender (NeurIPS 2025 · similarity 0.86)
- Self-EvolveRec: Self-Evolving Recommender Systems with LLM-based Directional Feedback (NeurIPS 2026 · similarity 0.86)
- Preference Diffusion for Recommendation (ICLR 2025 · similarity 0.86)
- Objective Shaping with Hard Negatives: Windowed Partial AUC Optimization for RL-based LLM Recommenders (NeurIPS 2026 · similarity 0.86)
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.
