Leveraging Memory Retrieval to Enhance LLM-based Generative Recommendation

TL;DR

This paper introduces AutoMR, a framework that enhances LLM-based generative recommendation by retrieving long-term user interests from memory, overcoming the limited context window of LLMs and improving recommendation quality.

Contribution

We propose AutoMR, a novel memory retrieval framework that enables LLMs to incorporate long-term user interests for better generative recommendations.

Findings

AutoMR significantly improves recommendation accuracy on real-world datasets.

Memory retrieval effectively captures long-term user interests.

AutoMR outperforms baseline models in experimental evaluations.

Abstract

Leveraging Large Language Models (LLMs) to harness user-item interaction histories for item generation has emerged as a promising paradigm in generative recommendation. However, the limited context window of LLMs often restricts them to focusing on recent user interactions only, leading to the neglect of long-term interests involved in the longer histories. To address this challenge, we propose a novel Automatic Memory-Retrieval framework (AutoMR), which is capable of storing long-term interests in the memory and extracting relevant information from it for next-item generation within LLMs. Extensive experimental results on two real-world datasets demonstrate the effectiveness of our proposed AutoMR framework in utilizing long-term interests for generative recommendation.