Training Large Recommendation Models via Graph-Language Token Alignment

TL;DR

This paper introduces a novel framework called Graph-Language Token Alignment (GLTA) that leverages large language models for recommendation systems by aligning graph nodes with LLM tokens, improving item prediction accuracy.

Contribution

The paper proposes a new method to integrate LLMs into recommender systems through token alignment and introduces Graph-Language Logits Matching for end-to-end item prediction.

Findings

GLTA outperforms baseline models on benchmark datasets.

Token alignment improves recommendation accuracy.

Ablation studies confirm the effectiveness of each component.

Abstract

Recommender systems (RS) have become essential tools for helping users efficiently navigate the overwhelming amount of information on e-commerce and social platforms. However, traditional RS relying on Collaborative Filtering (CF) struggles to integrate the rich semantic information from textual data. Meanwhile, large language models (LLMs) have shown promising results in natural language processing, but directly using LLMs for recommendation introduces challenges, such as ambiguity in generating item predictions and inefficiencies in scalability. In this paper, we propose a novel framework to train Large Recommendation models via Graph-Language Token Alignment. By aligning item and user nodes from the interaction graph with pretrained LLM tokens, GLTA effectively leverages the reasoning abilities of LLMs. Furthermore, we introduce Graph-Language Logits Matching (GLLM) to optimize token alignment for end-to-end item prediction, eliminating ambiguity in the free-form text as recommendation results. Extensive experiments on three benchmark datasets demonstrate the effectiveness of GLTA, with ablation studies validating each component.