Towards Next-Generation LLM-based Recommender Systems: A Survey and Beyond

The official GitHub page for the survey paper "Towards Next-Generation LLM-based Recommender Systems: A Survey and Beyond".

[arXiv: https://arxiv.org/abs/2410.19744]

Basic Information

Introduction

The comparison between this work and existing surveys

Paper	Non-Gen. RS	Gen. RS	Scen.	Aca.	Ind.	Pipeline	Highlights
'A survey on large language models for recommendation'	✅	✅	common (all kinds)	✅		(1) Discriminative LLM4REC (2) Generative LLM4REC Modeling Paradigms: (i) LLM Embeddings + RS (ii) LLM Tokens + RS (iii) LLM as RS	focuses on expanding the capacity of language models
'How Can Recommender Systems Benefit from Large Language Models: A Survey'	✅		common (all kinds)		✅	(1) Where to adapt to LLM (2) How to adapt to LLM	from the angle of the whole pipeline in industrial recommender systems
'A Survey on Large Language Models for Personalized and Explainable Recommendations'	✅		personalized and explainable RecSys	✅		(1) Explanation Generating for Recommendation	focuses on utilizing LLMs for personalized explanation generating task
'Recommender systems in the era of large language models (llms)'	✅		common (all kinds)	✅		(1) Pre-training (2) Fine-tuning (3) Prompting	comprehensively reviews such domain-specific techniques for adapting LLMs to recommendations
'A Review of Modern Recommender Systems Using Generative Models (Gen-RecSys)'		✅	(1) interaction-driven (2) text-driven (3) multimodal	✅		(1) Generative Models for Interaction-Driven Recommendation (2) Large Language Models in Recommendation (3) Generative Multimodal Recommendation Systems	aims to connect the key advancements in RS using Generative Models (Gen-RecSys)
Multimodal Pretraining, Adaptation, and Generation for Recommendation: A Survey	✅		multimodal recommendation	✅		(1) Multimodal Pretraining for Recommendation (2) Multimodal Adaption for Recommendation (3) Multimodal Generation for Recommendation	seeks to provide a comprehensive exploration of the latest advancements and future trajectories in multimodal pretraining, adaptation, and generation techniques, as well as their applications to recommender systems
'Large language models for generative recommendation: A survey and visionary discussions'		✅	common (all kinds)	✅		(1) ID Creation Methods (2) How to Do Generative Recommendation	reviews the recent progress of LLM-based generative recommendation and provides a general formulation for each generative recommendation task according to relevant research
Ours	✅	✅	common (all kinds)	✅	✅	(1) Representing and Understanding (2) Scheming and Utilizing (3) Industrial Deploying	(1) reviews existing works from the perspective of recommender system community (2) clearly discuss the gap from academic research to industrial application

(Gen.: Generative, RS: Recommendation System, Scen.: Scenarios, Aca.: Academic, Ind.: Industrial)

Structure

Pipeline

Different paradigm

LLM-using

Representing and Understanding

Representing

Uni-Modality

1. LLaRA: Large Language-Recommendation Assistant
2. DRDT: Dynamic Reflection with Divergent Thinking for LLM-based Sequential Recommendation
3. Modeling User Viewing Flow using Large Language Models for Article Recommendation
4. Harnessing Large Language Models for Text-Rich Sequential Recommendation
5. FineRec: Exploring Fine-grained Sequential Recommendation
6. Enhancing Sequential Recommendation via LLM-based Semantic Embedding Learning
7. A Multi-facet Paradigm to Bridge Large Language Model and Recommendation
8. Understanding Before Recommendation: Semantic Aspect-Aware Review Exploitation via Large Language Models
9. Representation Learning with Large Language Models for Recommendation
10. LLM-Enhanced User-Item Interactions: Leveraging Edge Information for Optimized Recommendations
11. GenRec: Large Language Model for Generative Recommendation
12. IDGenRec: LLM-RecSys Alignment with Textual ID Learning
13. Collaborative Large Language Model for Recommender Systems
14. Multiple Key-value Strategy in Recommendation Systems Incorporating Large Language Model

Multi-Modality

1. LLMRec: Large Language Models with Graph Augmentation for Recommendation
2. InteraRec: Interactive Recommendations Using Multimodal Large Language Models
3. Zero-Shot Recommendations with Pre-Trained Large Language Models for Multimodal Nudging
4. Large Language Models for Next Point-of-Interest Recommendation
5. Heterogeneous Knowledge Fusion: A Novel Approach for Personalized Recommendation via LLM
6. MMREC: LLM Based Multi-Modal Recommender System
7. Harnessing Multimodal Large Language Models for Multimodal Sequential Recommendation
8. X-Reflect: Cross-Reflection Prompting for Multimodal Recommendation

Understanding

Pre-Recommendation Explanations

1. Understanding Before Recommendation: Semantic Aspect-Aware Review Exploitation via Large Language Models
2. Learning Structure and Knowledge Aware Representation with Large Language Models for Concept Recommendation
3. LLM-Guided Multi-View Hypergraph Learning for Human-Centric Explainable Recommendation
4. RDRec: Rationale Distillation for LLM-based Recommendation
5. User-Centric Conversational Recommendation: Adapting the Need of User with Large Language Models
6. LLMRG: Improving Recommendations through Large Language Model Reasoning Graphs

Post-Recommendation Explanations

1. Unlocking the Potential of Large Language Models for Explainable Recommendations
2. Fine-Tuning Large Language Model Based Explainable Recommendation with Explainable Quality Reward
3. LLM4Vis: Explainable Visualization Recommendation using ChatGPT
4. Navigating User Experience of ChatGPT-based Conversational Recommender Systems: The Effects of Prompt Guidance and Recommendation Domain
5. DRE: Generating Recommendation Explanations by Aligning Large Language Models at Data-level
6. Uncertainty-Aware Explainable Recommendation with Large Language Models
7. Where to Move Next: Zero-shot Generalization of LLMs for Next POI Recommendation
8. Leveraging ChatGPT for Automated Human-centered Explanations in Recommender Systems
9. Logic-Scaffolding: Personalized Aspect-Instructed Recommendation Explanation Generation using LLMs
10. Leveraging Large Language Models in Conversational Recommender Systems
11. Chat-REC: Towards Interactive and Explainable LLMs-Augmented Recommender System
12. Leveraging Large Language Models for Recommendation and Explanation
13. GPT as a Baseline for Recommendation Explanation Texts
14. Supporting Student Decisions on Learning Recommendations: An LLM-Based Chatbot with Knowledge Graph Contextualization for Conversational Explainability and Mentoring
15. Knowledge Graphs as Context Sources for LLM-Based Explanations of Learning Recommendations
16. BookGPT: A General Framework for Book Recommendation Empowered by Large Language Model
17. LLMRec: Benchmarking Large Language Models on Recommendation Task
18. PAP-REC: Personalized Automatic Prompt for Recommendation Language Model
19. RecMind: Large Language Model Powered Agent For Recommendation
20. Prompt Distillation for Efficient LLM-based Recommendation

Scheming and Utilizing

Non-Generative LLM-based Approaches

LLM Retraing

Naive Fine-Tuning

1. Rethinking Large Language Model Architectures for Sequential Recommendations
2. LLMRec: Benchmarking Large Language Models on Recommendation Task
3. Improving Sequential Recommendations with LLMs
4. An Unified Search and Recommendation Foundation Model for Cold-Start Scenario
5. Recommender AI Agent: Integrating Large Language Models for Interactive Recommendations
6. Leveraging Large Language Models for Sequential Recommendation
7. A Large Language Model Enhanced Conversational Recommender System
8. Exploring Fine-tuning ChatGPT for News Recommendation
9. Aligning Large Language Models with Recommendation Knowledge
10. Conversational Recommender System and Large Language Model Are Made for Each Other in E-commerce Pre-sales Dialogue
11. Data-Efficient Fine-Tuning for LLM-based Recommendation
12. Large Language Model with Graph Convolution for Recommendation
13. A Bi-Step Grounding Paradigm for Large Language Models in Recommendation Systems
14. Learning Structure and Knowledge Aware Representation with Large Language Models for Concept Recommendation
15. LoRec: Large Language Model for Robust Sequential Recommendation against Poisoning Attacks
16. To Recommend or Not: Recommendability Identification in Conversations with Pre-trained Language Models
17. Large Language Models for Next Point-of-Interest Recommendation
18. Leveraging large language models in conversational recommender systems
19. Fine-Tuning Large Language Model Based Explainable Recommendation with Explainable Quality Reward
20. Enhancing Recommendation Diversity by Re-ranking with Large Language Models
21. NoteLLM: A Retrievable Large Language Model for Note Recommendation
22. Harnessing Large Language Models for Text-Rich Sequential Recommendation
23. LLM4DSR: Leveraing Large Language Model for Denoising Sequential Recommendation
24. Beyond Inter-Item Relations: Dynamic Adaptive Mixture-of-Experts for LLM-Based Sequential Recommendation
25. Improving Sequential Recommendations with LLMs

Instruction Tuning

1. A Multi-facet Paradigm to Bridge Large Language Model and Recommendation
2. Heterogeneous Knowledge Fusion: A Novel Approach for Personalized Recommendation via LLM
3. Exploring Large Language Model for Graph Data Understanding in Online Job Recommendations
4. Item-side Fairness of Large Language Model-based Recommendation System
5. Integrating Large Language Models into Recommendation via Mutual Augmentation and Adaptive Aggregation
6. RecRanker: Instruction Tuning Large Language Model as Ranker for Top-k Recommendation
7. E4SRec: An Elegant Effective Efficient Extensible Solution of Large Language Models for Sequential Recommendation
8. LLaRA: Large Language-Recommendation Assistant
9. Unlocking the Potential of Large Language Models for Explainable Recommendations
10. LlamaRec: Two-Stage Recommendation using Large Language Models for Ranking
11. Review-driven Personalized Preference Reasoning with Large Language Models for Recommendation

Low-Rank Adapataion (LoRA)

1. ONCE: Boosting Content-based Recommendation with Both Open- and Closed-source Large Language Models
2. Breaking the Length Barrier: LLM-Enhanced CTR Prediction in Long Textual User Behaviors
3. TALLRec: An Effective and Efficient Tuning Framework to Align Large Language Model with Recommendation
4. Harnessing Large Language Models for Text-Rich Sequential Recommendation
5. E4SRec: An Elegant Effective Efficient Extensible Solution of Large Language Models for Sequential Recommendation
6. Enhancing Content-based Recommendation via Large Language Model
7. Large Language Model Distilling Medication Recommendation Model
8. LLaRA: Large Language-Recommendation Assistant
9. Exact and Efficient Unlearning for Large Language Model-based Recommendation
10. Towards Efficient and Effective Unlearning of Large Language Models for Recommendation
11. LLM-based Federated Recommendation
12. Aligning Large Language Models for Controllable Recommendations
13. Lifelong Personalized Low-Rank Adaptation of Large Language Models for Recommendation
14. Harnessing Multimodal Large Language Models for Multimodal Sequential Recommendation
15. GANPrompt: Enhancing Robustness in LLM-Based Recommendations with GAN-Enhanced Diversity Prompts
16. DELRec: Distilling Sequential Pattern to Enhance LLM-based Recommendation
17. CoLLM: Integrating Collaborative Embeddings into Large Language Models for Recommendation

LLM Reusing

Direct Utilizing

1. Knowledge Plugins: Enhancing Large Language Models for Domain-Specific Recommendations
2. DRDT: Dynamic Reflection with Divergent Thinking for LLM-based Sequential Recommendation
3. Zero-Shot Next-Item Recommendation using Large Pretrained Language Models
4. Zero-Shot Recommendations with Pre-Trained Large Language Models for Multimodal Nudging
5. Large Language Models are Competitive Near Cold-start Recommenders for Language- and Item-based Preferences
6. Large Language Model Augmented Narrative Driven Recommendations
7. ChatGPT for Conversational Recommendation: Refining Recommendations by Reprompting with Feedback
8. Federated Recommendation via Hybrid Retrieval Augmented Generation
9. Re2LLM: Reflective Reinforcement Large Language Model for Session-based Recommendation
10. RecMind: Large Language Model Powered Agent For Recommendation
11. Reindex-Then-Adapt: Improving Large Language Models for Conversational Recommendation
12. Improve Temporal Awareness of LLMs for Sequential Recommendation
13. Tired of Plugins? Large Language Models Can Be End-To-End Recommenders
14. Where to Move Next: Zero-shot Generalization of LLMs for Next POI Recommendation
15. Large Language Models are Zero-Shot Rankers for Recommender Systems
16. InteraRec: Interactive Recommendations Using Multimodal Large Language Models
17. Large Language Models are Learnable Planners for Long-Term Recommendation
18. Logic-Scaffolding: Personalized Aspect-Instructed Recommendation Explanation Generation using LLMs
19. ReLLa: Retrieval-enhanced Large Language Models for Lifelong Sequential Behavior Comprehension in Recommendation
20. RDRec: Rationale Distillation for LLM-based Recommendation
21. LLM-Guided Multi-View Hypergraph Learning for Human-Centric Explainable Recommendation
22. Large Language Model Interaction Simulator for Cold-Start Item Recommendation
23. LLMRec: Large Language Models with Graph Augmentation for Recommendation
24. LKPNR: LLM and KG for Personalized News Recommendation Framework
25. LLM4Vis: Explainable Visualization Recommendation using ChatGPT
26. LLMRG: Improving Recommendations through Large Language Model Reasoning Graphs
27. Large Language Models for Intent-Driven Session Recommendations
28. Understanding Before Recommendation: Semantic Aspect-Aware Review Exploitation via Large Language Models
29. Leveraging Large Language Models (LLMs) to Empower Training-Free Dataset Condensation for Content-Based Recommendation
30. CoRAL: Collaborative Retrieval-Augmented Large Language Models Improve Long-tail Recommendation
31. Common Sense Enhanced Knowledge-based Recommendation with Large Language Model
32. DynLLM: When Large Language Models Meet Dynamic Graph Recommendation
33. FineRec: Exploring Fine-grained Sequential Recommendation
34. LLM4SBR: A Lightweight and Effective Framework for Integrating Large Language Models in Session-based Recommendation
35. Breaking the Barrier: Utilizing Large Language Models for Industrial Recommendation Systems through an Inferential Knowledge Graph
36. Sequential Recommendation with Latent Relations based on Large Language Model
37. News Recommendation with Category Description by a Large Language Model
38. PAP-REC: Personalized Automatic Prompt for Recommendation Language Model
39. Large Language Models as Data Augmenters for Cold-Start Item Recommendation
40. MMREC: LLM Based Multi-Modal Recommender System
41. DaRec: A Disentangled Alignment Framework for Large Language Model and Recommender System
42. X-Reflect: Cross-Reflection Prompting for Multimodal Recommendation
43. LLM4MSR: An LLM-Enhanced Paradigm for Multi-Scenario Recommendation
44. LLM-Based Aspect Augmentations for Recommendation Systems
45. Towards Open-World Recommendation with Knowledge Augmentation from Large Language Models
46. LLM-Rec: Personalized Recommendation via Prompting Large Language Models

Prompt Tuning

1. PMG: Personalized Multimodal Generation with Large Language Models
2. Language-Based User Profiles for Recommendation
3. Prompt Tuning Large Language Models on Personalized Aspect Extraction for Recommendations
4. RecPrompt: A Prompt Tuning Framework for News Recommendation Using Large Language Models

In-Context Learning (ICL)

1. Chat-REC: Towards Interactive and Explainable LLMs-Augmented Recommender System
2. DRE: Generating Recommendation Explanations by Aligning Large Language Models at Data-level
3. Efficient and Responsible Adaptation of Large Language Models for Robust Top-k Recommendations
4. Uncovering ChatGPT’s Capabilities in Recommender Systems
5. New Community Cold-Start Recommendation: A Novel Large Language Model-based Method
6. LANE: Logic Alignment of Non-tuning Large Language Models and Online Recommendation Systems for Explainable Reason Generation

Generative LLM-based Approaches

LLM Retraing

Naive Fine-Tuning

1. GPT4Rec: A Generative Framework for Personalized Recommendation and User Interests Interpretation
2. LLM-Enhanced User-Item Interactions: Leveraging Edge Information for Optimized Recommendations
3. RecGPT: Generative Personalized Prompts for Sequential Recommendation via ChatGPT Training Paradigm
4. RecGPT: Generative Pre-training for Text-based Recommendation
5. IDGenRec: LLM-RecSys Alignment with Textual ID Learning
6. Collaborative large language model for recommender systems
7. CALRec: Contrastive Alignment of Generative LLMs For Sequential Recommendation

Instruction Tuning

1. Multiple Key-Value Strategy in Recommendation Systems Incorporating Large Language Model

Low-Rank Adaptation (LoRA)

1. Genrec: Large Language Model for Generative Recommendation

LLM Reusing

Direct Utilizing

1. How to Index Item IDs for Recommendation Foundation Models
2. Supporting Student Decisions on Learning Recommendations: An LLM-based Chatbot with Knowledge Graph Contextualization for Conversational Explainability and Mentoring
3. Large Language Models as Zero-Shot Conversational Recommenders
4. Bookgpt: A General Framework for Book Recommendation Empowered by Large Language Model
5. ONCE: Boosting Content-based Recommendation with Both Open- and Closed-source Large Language Models

Prompt Tuning

1. PMG: Personalized Multimodal Generation with Large Language Models

Components or Strategies for Generative Recommendation

1. Learnable Tokenizer for LLM-based Generative Recommendation

The existing LLM-based works for generative recommendations

Model/Paper	Task/Domain	Data Modality	Main Techniques	Source Code
'Multiple Key-value Strategy in Recommendation Systems Incorporating Large Language Model'	sequential recommendation	multiple key-value data	pre-train, instruction tuning	~
'Large language models as zero-shot conversational recommenders'	zero-shot conversational recommendation	text (conversational recommendation dataset)	prompt	https://github.com/AaronHeee/LLMs-as-Zero-Shot-Conversational-RecSys
'Bookgpt: A general framework for book recommendation empowered by large language model'	book recommendation	interaction, text	prompt	https://github.com/zhiyulee-RUC/bookgpt
'How to index item ids for recommendation foundation models'	sequential recommendation	interaction, text	item ID indexing	https://github.com/Wenyueh/LLM-RecSys-ID
'Supporting student decisions on learning recommendations: An llm-based chatbot with knowledge graph contextualization for conversational explainability and mentoring'	learning recommendation	graph data, text	~	~
'GPT4Rec: A generative framework for personalized recommendation and user interests interpretation'	next-item prediction	interaction, item title	prompt, pre-train, fine-tune	~
'LLM-Enhanced User-Item Interactions: Leveraging Edge Information for Optimized Recommendations'	item recommendation	interaction	prompt, pre-train, fine-tune	https://github.com/anord-wang/LLM4REC.git
'RecGPT: Generative Personalized Prompts for Sequential Recommendation via ChatGPT Training Paradigm'	sequential recommendation	sequences of words	prompt, pre-train, fine-tune	~
'RecGPT: Generative Pre-training for Text-based Recommendation'	rating prediction, sequential recommendation	text	pre-train, fine-tune	https://github.com/VinAIResearch/RecGPT
'Genrec: Large language model for generative recommendation'	movie recommendation	interaction, textual-information	prompt, pre-train, fine-tune	https://github.com/rutgerswiselab/GenRec
'IDGenRec: LLM-RecSys Alignment with Textual ID Learning'	sequential recommendation, zero-shot recommendation	interaction, text	natural language generation	https://github.com/agiresearch/IDGenRec
'Collaborative large language model for recommender systems'	item recommendation	interaction, text	prompt, pre-train, fine-tune	https://github.com/yaochenzhu/llm4rec
'PMG: Personalized Multimodal Generation with Large Language Models'	personalized multimodal generation	text, image, audio, etc	prompt, pre-train, Prompt Tuning (P-Tuning V2)	https://github.com/mindspore-lab/models/tree/master/research/huawei-noah/PMG
'CALRec: Contrastive Alignment of Generative LLMs For Sequential Recommendation'	sequential recommendation	interaction, text	pre-train, fine-tune, contrastive learning	~
'Once: Boosting content-based recommendation with both open-and closed-source large language models'	content-based recommendation (news recommendation, book recommendation)	interaction, text	prompt	https://github.com/Jyonn/ONCE

Industrial Deploying

Large-Scale Industrial Scenarios

1. Breaking the Barrier: Utilizing Large Language Models for Industrial Recommendation Systems through an Inferential Knowledge Graph
2. A Large Language Model Enhanced Sequential Recommender for Joint Video and Comment Recommendation
3. RecGPT: Generative Personalized Prompts for Sequential Recommendation via ChatGPT Training Paradigm
4. Knowledge Adaptation from Large Language Model to Recommendation for Practical Industrial Application
5. Enhancing Sequential Recommendation via LLM-based Semantic Embedding Learning
6. Actions speak louder than words: Trillion-parameter sequential transducers for generative recommendations
7. Breaking the length barrier: LLM-Enhanced CTR Prediction in Long Textual User Behaviors
8. LLM4SBR: A Lightweight and Effective Framework for Integrating Large Language Models in Session-based Recommendation

Acceleration

1. A Decoding Acceleration Framework for Industrial Deployable LLM-based Recommender Systems

Cold Start

1. An Unified Search and Recommendation Foundation Model for Cold-Start Scenario
2. Knowledge Adaptation from Large Language Model to Recommendation for Practical Industrial Application
3. Breaking the Barrier: Utilizing Large Language Models for Industrial Recommendation Systems through an Inferential Knowledge Graph

Dynamic Update

1. DynLLM: When Large Language Models Meet Dynamic Graph Recommendation
2. Actions speak louder than words: Trillion-parameter sequential transducers for generative recommendations
3. COSMO: A large-scale e-commerce common sense knowledge generation and serving system at Amazon

Adapting to Business Customization Requirements

1. Modeling User Viewing Flow using Large Language Models for Article Recommendation
2. Ad Recommendation in a Collapsed and Entangled World
3. NoteLLM: A Retrievable Large Language Model for Note Recommendation
4. Beyond Labels: Leveraging Deep Learning and LLMs for Content Metadata
5. TRAWL: External Knowledge-Enhanced Recommendation with LLM Assistance
6. Heterogeneous Knowledge Fusion: A Novel Approach for Personalized Recommendation via LLM
7. COSMO: A large-scale e-commerce common sense knowledge generation and serving system at Amazon

The existing LLM-based works which could be deployed in industry

Model/Paper	Company	Task/Domain	Highlights
LLM-KERec, 'Breaking the Barrier: Utilizing Large Language Models for Industrial Recommendation Systems through an Inferential Knowledge Graph'	Ant Group	E-commerce Recommendation	Constructs a complementary knowledge graph by LLMs
LSVCR, 'A Large Language Model Enhanced Sequential Recommender for Joint Video and Comment Recommendation'	KuaiShou	Video Recommendation	Sequential recommendation model and supplemental LLM recommender
RecGPT, 'RecGPT: Generative Personalized Prompts for Sequential Recommendation via ChatGPT Training Paradigm'	KuaiShou	Sequential Recommendation	Models user behavior sequences using personalized prompts with ChatGPT.
SAID, 'Enhancing sequential recommendation via llm-based semantic embedding learning'	Ant Group	Sequential Recommendation	Explicitly learns Semantically Aligned item ID embeddings based on texts by utilizing LLMs.
BAHE, 'Breaking the length barrier: Llm-enhanced CTR prediction in long textual user behaviors'	Ant Group	CTR Prediction	Uses LLM's shallow layers for user behavior embeddings and deep layers for behavior interactions.
'LLM4SBR: A Lightweight and Effective Framework for Integrating Large Language Models in Session-based Recommendation'	HuaWei	Session-based Recommendation	In short sequence data, LLM can infer preferences directly leveraging its language understanding capability without fine-tuning
DARE, 'A Decoding Acceleration Framework for Industrial Deployable LLM-based Recommender Systems'	HuaWei	CTR prediction	Identifies the issue of inference efficiency during deploying LLM-based recommendations and introduces speculative decoding to accelerate recommendation knowledge generation.
'An Unified Search and Recommendation Foundation Model for Cold-Start Scenario'	Ant Group	Multi-domain Recommendation	LLM is applied to the S&R multi-domain foundation model to extract domain-invariant text features.
LEARN, 'Knowledge Adaptation from Large Language Model to Recommendation for Practical Industrial Application'	Kuaishou	Sequential Recommendation	Integrates the open-world knowledge encapsulated in LLMs into RS.
'DynLLM: When Large Language Models Meet Dynamic Graph Recommendation'	Alibaba	E-commerce Recommendations	Generates user profiles based on textual historical purchase records and obtaining users' embeddings by LLM.
HSTU, 'Actions speak louder than words: Trillion-parameter sequential transducers for generative recommendations'	Meta	Generating user action sequence	Explores the scaling laws of RS; Optimising model architecture to accelerate inference.
COSMO, 'COSMO: A large-scale e-commerce common sense knowledge generation and serving system at Amazon'	Amazon	Semantic relevance and session-based recommendation	Is the first industry-scale knowledge system that adopts LLM to construct high-quality knowledge graphs and serve online applications
SINGLE, 'Modeling User Viewing Flow using Large Language Models for Article Recommendation'	Taobao, Alibaba	Article Recommendation	Summarises long articles and the user constant preference from view history by gpt-3.5-turbo or ChatGLM-6B.
'Ad Recommendation in a Collapsed and Entangled World'	Tencent	Ad Recommendation	Obtains user or items embeddings by LLM.
NoteLLM, 'NoteLLM: A Retrievable Large Language Model for Note Recommendation'	xiaohongshu.com	Item-to-item Note Recommendation	Obtains article embeddings and generate hashtags/categories information by LLaMA-2.
Genre Spectrum, 'Beyond Labels: Leveraging Deep Learning and LLMs for Content Metadata'	tubi.tv	Movie & TV series Recommendation	Obtains content metadata embeddings by LLM.
TRAWL, 'TRAWL: External Knowledge-Enhanced Recommendation with LLM Assistance'	WeChat, Tencent	Article Recommendation	Uses Qwen1.5-7B extract knowledge from articles.
HKFR, 'Heterogeneous Knowledge Fusion: A Novel Approach for Personalized Recommendation via LLM'	Meituan	Catering Recommendation	Uses heterogeneous knowledge fusion for recommendations.

Challenges

Cite

If you compare with, build on, or use aspects of this work, please cite the following:

@article{wang2024towards,
  title={Towards Next-Generation LLM-based Recommender Systems: A Survey and Beyond},
  author={Wang, Qi and Li, Jindong and Wang, Shiqi and Xing, Qianli and Niu, Runliang and Kong, He and Li, Rui and Long, Guodong and Chang, Yi and Zhang, Chengqi},
  journal={arXiv preprint arXiv:2410.19744},
  year={2024}
}