RecAI: Leveraging Large Language Models for Next-Generation Recommender Systems

Jianxun Lian,Yuxuan Lei,Xu Huang,Jing Yao,Wei Xu,Xing Xie

DOI: https://doi.org/10.1145/3589335.3651242

2024-03-11

Abstract:This paper introduces RecAI, a practical toolkit designed to augment or even revolutionize recommender systems with the advanced capabilities of Large Language Models (LLMs). RecAI provides a suite of tools, including Recommender AI Agent, Recommendation-oriented Language Models, Knowledge Plugin, RecExplainer, and Evaluator, to facilitate the integration of LLMs into recommender systems from multifaceted perspectives. The new generation of recommender systems, empowered by LLMs, are expected to be more versatile, explainable, conversational, and controllable, paving the way for more intelligent and user-centric recommendation experiences. We hope the open-source of RecAI can help accelerate evolution of new advanced recommender systems. The source code of RecAI is available at \url{<a class="link-external link-https" href="https://github.com/microsoft/RecAI" rel="external noopener nofollow">this https URL</a>}.

Information Retrieval,Artificial Intelligence

What problem does this paper attempt to address?

This paper aims to explore how to utilize large - language models (LLMs) to enhance the capabilities of recommendation systems (RSs). Specifically, the paper proposes a toolkit named RecAI, which aims to enhance or even revolutionize recommendation systems by integrating LLMs. RecAI provides a series of tools, including Recommender AI Agent, Recommendation - oriented Language Models, Knowledge Plugin, RecExplainer and Evaluator, to promote the integration of LLMs in recommendation systems from multiple perspectives.

The paper points out that although LLMs have powerful natural - language - processing capabilities and can be directly used for a variety of tasks, it is not practical to directly use them as recommendation models. This is because the knowledge boundaries of LLMs are limited to the information at the time of their last training update and may not fully capture the item catalogs and properties in specific recommendation scenarios. In addition, user preference patterns are not only domain - specific but also change rapidly. Traditional recommendation models need to be frequently retrained or fine - tuned to capture these unique and constantly changing patterns, which are different from the general world knowledge encoded in LLMs.

Therefore, the paper proposes to overcome these challenges through the RecAI toolkit, making the new - generation recommendation systems more intelligent, multi - purpose, interpretable, conversation - friendly and controllable, thereby providing users with a more intelligent and user - centered recommendation experience.