This is the collection of the papers on LLM-Enhanced Recommender Systems (LLMERS).
Due to the latency requirements by the online recommenders, LLMERS has gained considerable attention recently. We would like to give the definition of the LLMERS in our survey.
💡 Definition: The conventional recommender systems are enhanced by large language models via assistance in training or supplementary for data, while no need for LLM inference during the service
⭐️ Category: LLMERS augments the conventional RS from its basic components, i.e., interaction data and recommendation model, so that we categorize them into three lines: Knowledge Enhancement, Interaction Enhancement and Model Enhancement.
🔥 [2024.12] We have released the survey and repo!
🔥 [2025.3] We have updated recent papers to v.2.
Beside, we also conclude existing surveys about Large Language Model for Recommender Systems as follows.