Research map / Large language models / Alignment and preferences
Mitigating Reward Hacking in LLM-based Recommendation: A Preference Optimization Approach
Heyu Chen, Junkang Wu, Guoqing Hu, Kexin Huang, Xiang Wang, Jiancan Wu
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 SIRIUS pseudo-negative | SimPO w/o LN · LastFM | HitRatio@1: 0.6633 -> 0.7315 | Improved |
| add SIRIUS pseudo-negative | SimPO w/o LN · Goodreads | HitRatio@1: 0.7010 -> 0.7857 | Improved |
| add SIRIUS pseudo-negative | SimPO w/o LN · Steam | HitRatio@1: 0.8002 -> 0.8550 | Improved |
| add SIRIUS pseudo-negative | DPO · LastFM | HitRatio@1: 0.6393 -> 0.6593 | Improved |
| add SIRIUS pseudo-negative | DPO · Goodreads | HitRatio@1: 0.6462 -> 0.6744 | Improved |
| add SIRIUS pseudo-negative | DPO · Steam | HitRatio@1: 0.7333 -> 0.7757 | Improved |
| add SIRIUS to DPO instead of using S-DPO with three negatives | DPO · Goodreads | HitRatio@1: 0.6744 -> 0.6744 | No clear effect |
| use DPO + SIRIUS instead of S-DPO with multiple negatives | DPO · Steam | HitRatio@1: 0.7712 -> 0.7757 | Improved |
| remove length normalization from SimPO (ablation) | SimPO · LastFM | HitRatio@1: 0.5928 -> 0.6633 | Improved |
| add SIRIUS to S-DPO-style training (SIRIUS runs keep improving past epoch 3) | S-DPO · LastFM, Goodreads, Steam | HitRatio@1 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- SimPO — SimPO: Simple Preference Optimization with a Reference-Free Reward (NeurIPS 2024) · better on 4 of 4 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Language Ranker: A Lightweight Ranking framework for LLM Decoding (NeurIPS 2025 · similarity 0.89)
- Objective Shaping with Hard Negatives: Windowed Partial AUC Optimization for RL-based LLM Recommenders (NeurIPS 2026 · similarity 0.88)
- On Softmax Direct Preference Optimization for Recommendation (NeurIPS 2024 · similarity 0.88)
- Mult-DPO: Multinomial Direct Preference Optimization for Recommender Systems (NeurIPS 2026 · similarity 0.88)
- Sail into the Headwind: Alignment via Robust Rewards and Dynamic Labels against Reward Hacking (ICLR 2025 · similarity 0.88)
- RRM: Robust Reward Model Training Mitigates Reward Hacking (ICLR 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.
