LWGR: Lagrangian-Constrained Personalized World Knowledge for Generative Recommendation
sigir26-153 · Full research paper · Lingyu Mu, Hao Deng, Haibo Xing, Kaican Lin, Zhitong Zhu, Zhengxiao Liu, Zheng Lin, Xiaoyi Zeng, Yu Zhang, Jinxin Hu
Generative Retrieval and Recommendat · 25% Knowledge Infusion and Open-World Au · 25% Generative Recommendation and Item T · 25% LLM-Derived Knowledge Infusion · 25%
Abstract
Recent progress in large language model (LLM) based generative recommendation (GR) shows that leveraging LLM world knowledge can substantially improve performance. However, existing methods rely on fixed, manually designed instructions to generate semantic knowledge and directly incorporate it into GR, which has two limitations: (1) fixed instructions cannot capture the multidimensional heterogeneity of user interests; (2) uncontrollable knowledge fusion may conflict with behavioral signals and harm recommendations. To address these limitations, we propose LWGR, a framework that leverages Lagrangian constraints to transfer users' personalized World knowledge from LLMs into Generative Recommendation. LWGR enhances GR along two axes: knowledge extraction and fusion. It builds user personalized soft instructions to extract behavior-relevant LLM world knowledge. Then, it formulates knowledge fusion as an optimization problem with explicitly bounded performance degradation, solved via a Lagrangian primal–dual method that selectively incorporates beneficial knowledge. We further design two training strategies for different LLM scales and a deployment scheme that combines nearline precomputation with lightweight online serving. Experiments on multiple public datasets and one industrial dataset show that LWGR outperforms eight state-of-the-art baselines by up to 11.23% and brings a 1.35% revenue lift on a large-scale advertising platform, demonstrating its effectiveness and practicality.
Assigned reviewers
| # | Reviewer | Match | Why (closest evidence publications in the taxonomy) | |
|---|---|---|---|---|
| 1 | Bo Chen 0023 | 63 | 76 Towards Open-World Recommendation with Knowledge Augmentation from Large Language Models (2024) Knowledge Infusion and Open-World Augmentation 76 Efficient and Deployable Knowledge Infusion for Open-World Recommendations via Large Language Models (2026) Knowledge Infusion and Open-World Augmentation 66 Large language models make sample-efficient recommender systems (2025) Efficiency and Deployment |
|
| 2 | Jieming Zhu | 59 | 76 Efficient and Deployable Knowledge Infusion for Open-World Recommendations via Large Language Models (2026) Knowledge Infusion and Open-World Augmentation 76 Towards Open-World Recommendation with Knowledge Augmentation from Large Language Models (2024) Knowledge Infusion and Open-World Augmentation 74 EAGER-LLM: Enhancing Large Language Models as Recommenders through Exogenous Behavior-Semantic Integration (2025) Generative Recommendation and Item Tokenization |
|
| 3 | Qijiong Liu | 55 | 78 ONCE: Boosting Content-based Recommendation with Both Open- and Closed-source Large Language Models (2024) LLM-Derived Knowledge Infusion 77 RecBase: Generative Foundation Model Pretraining for Zero-Shot Recommendation (2025) Generative Retrieval and Recommendation Architectures 59 Can LLMs Outshine Conventional Recommenders? A Comparative Evaluation (2025) Efficiency and Deployment |
Area chair
| Area chair | Match | |
|---|---|---|
| Pengyue Jia | 49 |
Area-chair candidates (7)
| Pengyue Jia assigned | 49 | 10 papers | |
| Kun Gai COI: co-author | 49 | 5 papers | |
| Xiaoyi Zeng COI: author | 48 | 11 papers | |
| Maolin Wang 0001 | 47 | 15 papers | |
| Weiming Liu 0005 | 47 | 8 papers | |
| Leyu Lin | 47 | 15 papers | |
| Jun Xu 0001 | 47 | 2 papers |
Candidates from the panel ranked by match score
| # | Reviewer | Match | Load | Why | |
|---|---|---|---|---|---|
| 1 | Bo Chen 0023 assigned | 63 | 3 | 76 Towards Open-World Recommendation with Knowledge Augmentation from Large Language Models (2024) Knowledge Infusion and Open-World Augmentation 76 Efficient and Deployable Knowledge Infusion for Open-World Recommendations via Large Language Models (2026) Knowledge Infusion and Open-World Augmentation |
|
| 2 | Jieming Zhu assigned | 59 | 3 | 76 Efficient and Deployable Knowledge Infusion for Open-World Recommendations via Large Language Models (2026) Knowledge Infusion and Open-World Augmentation 76 Towards Open-World Recommendation with Knowledge Augmentation from Large Language Models (2024) Knowledge Infusion and Open-World Augmentation |
|
| 3 | Bowen Zheng 0005 | 56 | 3 | 61 Generative Recommender with End-to-End Learnable Item Tokenization (2025) Item Tokenization and Semantic IDs 59 Adapting Large Language Models by Integrating Collaborative Semantics for Recommendation (2024) Generative Recommendation and Item Tokenization |
|
| 4 | Qijiong Liu assigned | 55 | 3 | 78 ONCE: Boosting Content-based Recommendation with Both Open- and Closed-source Large Language Models (2024) LLM-Derived Knowledge Infusion 77 RecBase: Generative Foundation Model Pretraining for Zero-Shot Recommendation (2025) Generative Retrieval and Recommendation Architectures |
|
| 5 | Zhengyi Yang 0007 | 55 | 3 | 64 A Bi-Step Grounding Paradigm for Large Language Models in Recommendation Systems (2025) Generative Recommendation and Item Tokenization 59 Large Language Model Can Interpret Latent Space of Sequential Recommender (2026) Generative Recommendation and Item Tokenization |
|
| 6 | Yang Zhang 0072 | 55 | 3 | 64 IGD: Token Decisiveness Modeling via Information Gain in LLMs for Personalized Recommendation (2025) Generative Recommendation and Item Tokenization 64 Decoding Matters: Addressing Amplification Bias and Homogeneity Issue in Recommendations for Large Language Models (2024) Generative Recommendation and Item Tokenization |
|
| 7 | Xuanhua Yang | 54 | 3 | 68 Breaking the Relevance-Diversity Seesaw: Hierarchical LLM Reasoning with RL for Industrial Novelty Recommendation (2026) Preference Alignment and Optimization 66 GenRec: A Preference-Oriented Generative Framework for Large-Scale Recommendation (2026) Generative Retrieval and Ranking |
|
| 8 | Huifeng Guo | 54 | 3 | 65 LLMTreeRec: Unleashing the Power of Large Language Models for Cold-Start Recommendations (2025) Generative Recommendation and Item Tokenization 63 SELF: Surrogate-light Feature Selection with Large Language Models in Deep Recommender Systems (2025) LLM-Derived Knowledge Infusion |
|
| 9 | Yupeng Hou | 53 | 3 | 67 Large Language Models are Zero-Shot Rankers for Recommender Systems (2024) Efficiency and Deployment 63 Inductive Generative Recommendation via Retrieval-based Speculation (2026) Item Tokenization and Semantic IDs |
|
| 10 | Jizhi Zhang | 53 | 3 | 64 Decoding Matters: Addressing Amplification Bias and Homogeneity Issue in Recommendations for Large Language Models (2024) Generative Recommendation and Item Tokenization 62 Learnable Item Tokenization for Generative Recommendation (2024) Item Tokenization and Semantic IDs |
|
| 11 | Shuchang Liu 0001 | 52 | 3 | 75 DLCRec: A Novel Approach for Managing Diversity in LLM-Based Recommender Systems (2025) Preference Alignment and Optimization 74 Recommendation as Language Processing (RLP): A Unified Pretrain, Personalized Prompt & Predict Paradigm (P5) (2022) Knowledge Infusion and Open-World Augmentation |
|
| 12 | Chongming Gao | 51 | 3 | 63 MGFRec: Towards Reinforced Reasoning Recommendation with Multiple Groundings and Feedback (2026) Preference Alignment and Optimization 63 DLCRec: A Novel Approach for Managing Diversity in LLM-Based Recommender Systems (2025) Preference Alignment and Optimization |
|
| 13 | Han Li 0005 | 51 | 3 | 63 OneSug: The Unified End-to-End Generative Framework for E-commerce Query Suggestion (2026) Recommendation with Search and Ranking 57 Towards Context-aware Reasoning-enhanced Generative Searching in E-commerce (2026) Recommendation with Search and Ranking |
|
| 14 | Qiang Luo 0004 | 50 | 3 | 61 DiffGRM: Diffusion-based Generative Recommendation Model (2026) Diffusion and Flow Recommenders 55 OneLoc: Geo-Aware Generative Recommender Systems for Local Life Service (2026) Item Tokenization and Semantic IDs |
|
| 15 | Junwei Pan | 50 | 3 | 77 Cross-Scale Collaboration between LLMs and Lightweight Sequential Recommenders with Domain-Specific Latent Reasoning (2026) Generative Recommendation and Item Tokenization 63 Empowering Large Language Model for Sequential Recommendation via Multimodal Embeddings and Semantic IDs (2025) Generative Recommendation and Item Tokenization |
|
| 16 | Jianling Wang | 50 | 3 | 77 Fresh Content Recommendation at Scale: A Multi-funnel Solution and the Potential of LLMs (2024) Efficiency and Deployment 55 LLMs for User Interest Exploration in Large-scale Recommendation Systems (2024) Sequential and Session-based Recommendation |
|
| 17 | Hao Deng 0011 COI: author | 50 | 3 | 60 RCLRec: Reverse Curriculum Learning for Modeling Sparse Conversions in Generative Recommendation (2026) Generative Retrieval and Ranking 58 REG4Rec: Reasoning-Enhanced Generative Model for Large-Scale Recommendation Systems (2026) Item Tokenization and Semantic IDs |
|
| 18 | Wei Guo 0006 | 50 | 3 | 66 Can LLMs Outshine Conventional Recommenders? A Comparative Evaluation (2025) Efficiency and Deployment 62 Killing Two Birds with One Stone: Unifying Retrieval and Ranking with a Single Generative Recommendation Model (2025) Generative Retrieval and Ranking |
|
| 19 | Yongqi Li 0001 | 49 | 3 | 75 Bridging Items and Language: A Transition Paradigm for Large Language Model-Based Recommendation (2024) Generative Recommendation and Item Tokenization 62 Data-efficient Fine-tuning for LLM-based Recommendation (2024) Efficiency and Deployment |
|
| 20 | Yang Song 0008 | 49 | 3 | 62 Large Language Models Enhanced Collaborative Filtering (2024) LLM-Enhanced Collaborative Filtering 59 GenSAR: Unifying Balanced Search and Recommendation with Generative Retrieval (2025) Recommendation with Search and Ranking |
|
| 21 | Ed H. Chi | 49 | 3 | 65 PLUM: Adapting Pre-trained Language Models for Industrial-scale Generative Recommendations (2026) Item Tokenization and Semantic IDs 60 LLMs for User Interest Exploration in Large-scale Recommendation Systems (2024) Generative Recommendation and Item Tokenization |
|
| 24 | Xiao Zhang 0034 | 49 | 3 | 63 Large Language Models Enhanced Collaborative Filtering (2024) Preference Alignment and Optimization 59 GenSAR: Unifying Balanced Search and Recommendation with Generative Retrieval (2025) Recommendation with Search and Ranking |
|
| 25 | Lichan Hong | 49 | 3 | 64 PLUM: Adapting Pre-trained Language Models for Industrial-scale Generative Recommendations (2026) Generative Recommendation and Item Tokenization 61 Recommender Systems with Generative Retrieval (2023) Item Tokenization and Semantic IDs |
|
| 26 | Hao Liu 0078 | 48 | 3 | 55 WeaveRec: An LLM-Based Cross-Domain Sequential Recommendation Framework with Model Merging (2026) Efficiency and Deployment 46 Mitigating Distribution Shifts in Sequential Recommendation: An Invariance Perspective (2025) Contrastive and Self-Supervised Sequential Recommendation |
|
| 28 | Bo Peng 0009 | 48 | 3 | 52 eCeLLM: Generalizing Large Language Models for E-commerce from Large-scale, High-quality Instruction Data (2024) Efficiency and Deployment 50 M2: Mixed Models With Preferences, Popularities and Transitions for Next-Basket Recommendation (2023) Efficient and Long-Sequence Sequential Recommendation |
|
| 29 | Lantao Hu | 48 | 3 | 51 Generative Recommender with End-to-End Learnable Item Tokenization (2025) Item Tokenization and Semantic IDs 50 Comprehensive List Generation for Multi-Generator Reranking (2025) Recommendation with Search and Ranking |
|
| 30 | Jiangxia Cao | 47 | 3 | 54 Enhancing Content-based Recommendation via Large Language Model (2024) Collaborative-Semantic Representation Alignment 50 CDSRNP: Cross-Domain Sequential Recommendation via Neural Process (2025) Cross-Domain Sequential Recommendation |
|
| 31 | Jinxin Hu COI: author | 47 | 3 | 64 REG4Rec: Reasoning-Enhanced Generative Model for Large-Scale Recommendation Systems (2026) Item Tokenization and Semantic IDs 60 RCLRec: Reverse Curriculum Learning for Modeling Sparse Conversions in Generative Recommendation (2026) Generative Retrieval and Ranking |
|
| 33 | Yongxin Ni | 47 | 3 | 50 An Image Dataset for Benchmarking Recommender Systems with Raw Pixels (2024) Recommendation Systems and Evaluation 50 Where to Go Next for Recommender Systems? ID- vs. Modality-based Recommender Models Revisited (2023) Multimodal Generative Recommendation |
|
| 34 | Jiancan Wu | 47 | 3 | 52 Customizing Language Models with Instance-wise LoRA for Sequential Recommendation (2024) Sequential and Session-based Recommendation 51 LLaRA: Large Language-Recommendation Assistant (2024) Sequential and Session-based Recommendation |