Research map / Large language models / Language, safety and interpretability
MoMoREC: A Multi-agent Motivation Generation Framework for Residual Semantic ID-Aware Recommendation
Yige Wang, Mingming Li, Li Wang, Kaichen Zhao, Wangming Li, Weipeng Jiang, Xueying LI
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 |
|---|---|---|---|
| use MoMoREC motivation semantic IDs instead of baseline SASRec | SASRec · Amazon Beauty | Recall@10: 0.0303 -> 0.1725 | Improved |
| use MoMoREC instead of baseline SASRec | SASRec · Amazon Video Game | Recall@10: 0.1425 -> 0.6578 | Improved |
| use MoMoREC instead of baseline SASRec | SASRec · Amazon Baby Product | Recall@10: 0.0624 -> 0.7285 | Improved |
| add Motivation Spreading to long-tail items (ablation) | MoMoREC · Amazon Beauty | Recall@10: 0.0791 -> 0.1725 | Improved |
| add residual semantic ID retrieval (vs cluster-based IDs only) (ablation) | MoMoREC · Amazon Beauty | Recall@10: 0.0942 -> 0.1725 | Improved |
| use LLM embeddings to initialize ID embeddings (LLMInit) instead of base | SASRec · Amazon Video Game | Recall@10: 0.1425 -> 0.0371 | Hurt |
| fine-tune the embedding model on motivation data vs unfine-tuned gte-Qwen2 (ablation) | embedding model · Embedding quality | Spearman: 0.8027 -> 0.8450 | Improved |
| use stronger LLM (gpt-4o-mini vs Qwen2.5-7B) as judging agent (ablation) | Judgment Agent · Human annotation agreement | agreement rate: 49.85% -> 72.48% | Improved |
| deploy motivation semantic ID in online model | Taobao online model · Taobao 88VIP A/B | UCTR / TCR / GMV: UCTR -1.3%, TCR +1%, GMV +6.3% | Mixed |
| add motivation semantic ID embeddings to SASRec | SASRec · Amazon Beauty | inference time: 550.00 -> 593.54 ms (+7.9%) | Hurt |
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.
- LLMESR — LLM-ESR: Large Language Models Enhancement for Long-tailed Sequential Recommendation (NeurIPS 2024) · better on 1 of 1 shared result
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
- LLM-ESR: Large Language Models Enhancement for Long-tailed Sequential Recommendation (NeurIPS 2024 · similarity 0.89)
- Semantic Convergence: Harmonizing Recommender Systems via Two-Stage Alignment and Behavioral Semantic Tokenization (AAAI 2025 · similarity 0.88)
- LLMEmb: Large Language Model Can Be a Good Embedding Generator for Sequential Recommendation (AAAI 2025 · similarity 0.88)
- Can LLMs Outshine Conventional Recommenders? A Comparative Evaluation (NeurIPS 2025 · similarity 0.88)
- FACE: A General Framework for Mapping Collaborative Filtering Embeddings into LLM Tokens (NeurIPS 2025 · similarity 0.88)
- Language Representations Can be What Recommenders Need: Findings and Potentials (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.
