Large Language Models Meet Collaborative Filtering: an Efficient All-round LLM-based Recommender System
KDD 2024(2024)
KAIST
Abstract
Collaborative filtering recommender systems (CF-RecSys) have shown successiveresults in enhancing the user experience on social media and e-commerceplatforms. However, as CF-RecSys struggles under cold scenarios with sparseuser-item interactions, recent strategies have focused on leveraging modalityinformation of user/items (e.g., text or images) based on pre-trained modalityencoders and Large Language Models (LLMs). Despite their effectiveness undercold scenarios, we observe that they underperform simple traditionalcollaborative filtering models under warm scenarios due to the lack ofcollaborative knowledge. In this work, we propose an efficient All-roundLLM-based Recommender system, called A-LLMRec, that excels not only in the coldscenario but also in the warm scenario. Our main idea is to enable an LLM todirectly leverage the collaborative knowledge contained in a pre-trainedstate-of-the-art CF-RecSys so that the emergent ability of the LLM as well asthe high-quality user/item embeddings that are already trained by thestate-of-the-art CF-RecSys can be jointly exploited. This approach yields twoadvantages: (1) model-agnostic, allowing for integration with various existingCF-RecSys, and (2) efficiency, eliminating the extensive fine-tuning typicallyrequired for LLM-based recommenders. Our extensive experiments on variousreal-world datasets demonstrate the superiority of A-LLMRec in variousscenarios, including cold/warm, few-shot, cold user, and cross-domainscenarios. Beyond the recommendation task, we also show the potential ofA-LLMRec in generating natural language outputs based on the understanding ofthe collaborative knowledge by performing a favorite genre prediction task. Ourcode is available at https://github.com/ghdtjr/A-LLMRec .
MoreTranslated text
Key words
Language Modeling,Statistical Language Modeling,Topic Modeling,Part-of-Speech Tagging,Syntax-based Translation Models
PDF
View via Publisher
AI Read Science
Must-Reading Tree
Example

Generate MRT to find the research sequence of this paper
Data Disclaimer
The page data are from open Internet sources, cooperative publishers and automatic analysis results through AI technology. We do not make any commitments and guarantees for the validity, accuracy, correctness, reliability, completeness and timeliness of the page data. If you have any questions, please contact us by email: report@aminer.cn
Chat Paper
Summary is being generated by the instructions you defined