Abstract
Due to the brilliant abilities in reasoning and understanding, the Large Language Model (LLM) has revolutionized the pattern of several fields, including recommender systems (RS). There have been a handful of research that focuses on empowering the RS by LLM. Recently, considering the latency and memory costs in real-world applications, LLM-Enhanced RS (LLMERS) is highlighted. This direction pushes the LLM into the online system with a large step by eliminating the utilization of LLM during inference. However, it is a rather cutting-edge field, and a comprehensive tutorial is desirable to conclude this direction.
In this tutorial, we investigate the most up-to-date works of LLM-enhanced RS to boost this direction. Based on the component of an RS model that the LLM aims to augment, the basic taxonomy includes Knowledge Enhancement, Interaction Enhancement and Model Enhancement. In each basic cluster, we identify and summarize what and how to augment the traditional RS. Besides, for further advancements in both industrial and academic fields, we give out its applications and several promising directions. A repository is released to facilitate access to the surveyed papers.
Survey Paper
Slides
KDD2025-LLMERS-Tutorial-Slides
Video
Comming soon …
Repository
Targe Audiance
To boost audience engagement during the tutorial, we will incorporate various interactive activities, such as live polls, real-time quizzes, and dedicated Q&A sessions, to encourage active participation. We will also use visually appealing and easy-to-follow examples to simplify complex topics, ensuring they are both understandable and relatable. Furthermore, we plan to create opportunities for group discussions, allowing attendees to share insights with peers. By fostering a dynamic and inclusive environment, we aim to make the tutorial both informative and engaging for all participants.
Tutorial Outline
- S1. Introduction to LLM for RS (20 mins)
- LLM as Recommender Systems
- LLM enhanced Recommender Systems
- S2. Knowledge Enhancement (30 mins)
- Introduction to Knowledge Enhancement
- Summary Text
- Knowledge Graph
- Combination
- S3. Interaction Enhancement (30 mins)
- Introduction to Interaction Enhancement
- Text-based Method
- Score-based Method
- Coffee Break (30 mins)
- S4. Model Enhancement 1 (35 mins)
- Introduction to Model Enhancement
- Model Initialization
- Model Distillation
- S5. Model Enhancement 2 (35 mins)
- Embedding Utilization
- Embedding Guidance
- S6. Applications, Trends, and Conclusion (10 mins)
- Applications
- Trends and Future Directions
- Conclusion
Presenter
Mr. Yuhao Wang is a Ph.D. candidate at the City University of Hong Kong (CityU). He works in recommender systems, especially multi-scenario and multi-task modeling.
Mr. Yejing Wang is a Ph.D. candidate at the City University of Hong Kong (CityU). His research interest lies in Recommender Systems and the Large Language Model.
Dr. Zijian Zhang is an assistant professor at Jilin University (JLU), working on recommender systems, urban computing, and Large Language Models.
Mr. Maolin Wang is a Ph.D. candidate at the City University of Hong Kong (CityU), working on information retrieval and large language models.
Mr. Pengyue Jia is a Ph.D. candidate at the City University of Hong Kong (CityU). His research interest lies in recommender systems, information retrieval, and GeoAI.
Mr. Ziwei Liu is a Ph.D. candidate at the City University of Hong Kong (CityU). His research interest lies in recommender systems and Large Language Models.
Contributor
Dr. Qidong Liu is a joint-PhD with the Xi’an Jiaotong University (XJTU) and City University of Hong Kong (CityU).His research interests include Recommender Systems and the Large Language Model.He has published more than 20 papers in top conferences and journals (e.g. SIGIR, NeurIPS, AAAI, TOIS). Please find more information at https://liuqidong07.github.io/. Dr. Liu is a tutorial organizer at KDD’23, WWW’23, and IJCAI’23.
Prof. Xiangyu Zhao is an assistant professor at the City University of Hong Kong (CityU). His research interests include data mining and machine learning, especially (1) Personalization, Recommender Systems, Online Advertising, and Search Engines; and (2) Joint Modeling, AutoML, Reinforcement Learning, and Multimodal. He has published over 100 papers in top conferences and journals (e.g. KDD, SIGIR, RecSys, WWW, TKDE, TOIS). His research has been awarded ICDM’22 and ICDM’21 Best-ranked Papers, Global Top 25 Chinese New Stars in AI (Data Mining), CCF-Tencent Open Fund (twice), CCF-Ant Research Fund, Criteo Faculty Research Award, and nomination for Joint AAAI/ACM SIGAI Doctoral Dissertation Award. Please find more information at https://zhaoxyai.github.io/. Dr. Zhao is a lead tutor at KDD’23, WWW’21/22/23, IJCAI’21/23, and WSDM’23.
Ms. Yuqi Sun is a master student at Xi’an Jiaotong University (XJTU). Her research interests include recommender systems and the large language model.
Mr. Xiang Li is a researcher with the 2012 Laboratory, Huawei. His current research interests include artificial intelligence.
Dr. Chong Chen is a senior researcher at Huawei Inc. His research interests are mainly in LLM application methods, such as Data-centric LLM, RAG, Agent, NL2SQL, Search, Recommendation and etc. He has published more than 30 papers in top conferences and journals such as WWW, SIGIR, AAAI and TOIS.
Mr. Wei Huang graduated from Xiamen University. He is currently a principal engineer, who is researching LLM.
Prof. Feng Tian is a Professor at Xi’an Jiaotong University (XJTU). His research interests include smart education, especially (1) Personalized Intelligent Tutor Systems and (2) Online Educational Data Mining. He has published over 100 papers in top conferences and journals (e.g. TKDE, ICDE, AAAI, EMNLP and TMM). He has been awarded the China International Science and Technology Cooperation Award, CCF Science and Technology Progress Award, etc.
Citation
Welcome to cite our survey paper!
1 | @article{liu2024large, |