EmbSum: Leveraging the Summarization Capabilities of Large Language Models for Content-Based Recommendations

TL;DR

EmbSum is a new framework that uses large language models to generate user interest summaries and improve content-based recommendations by capturing long user engagement histories and calculating relevance scores.

Contribution

It introduces EmbSum, a novel method leveraging pretrained LLMs and poly-attention layers for offline user and item embedding computation, surpassing state-of-the-art accuracy.

Findings

Outperforms existing methods on multiple datasets.

Uses fewer parameters than comparable models.

Generates meaningful user interest summaries.

Abstract

Content-based recommendation systems play a crucial role in delivering personalized content to users in the digital world. In this work, we introduce EmbSum, a novel framework that enables offline pre-computations of users and candidate items while capturing the interactions within the user engagement history. By utilizing the pretrained encoder-decoder model and poly-attention layers, EmbSum derives User Poly-Embedding (UPE) and Content Poly-Embedding (CPE) to calculate relevance scores between users and candidate items. EmbSum actively learns the long user engagement histories by generating user-interest summary with supervision from large language model (LLM). The effectiveness of EmbSum is validated on two datasets from different domains, surpassing state-of-the-art (SoTA) methods with higher accuracy and fewer parameters. Additionally, the model's ability to generate summaries of user interests serves as a valuable by-product, enhancing its usefulness for personalized content recommendations.