Large Language Models Make Sample-Efficient Recommender Systems
Jianghao Lin, Xinyi Dai, Rong Shan, Bo Chen, Ruiming Tang, Yong Yu, Weinan Zhang
TL;DR
The paper addresses data sparsity in recommender systems by introducing Laser, a framework that leverages large language models in two ways: (i) using LLMs as direct CTR predictors (Laser_LLM_only) and (ii) using LLMs to generate textual features for traditional CRMs (Laser_LLM+CRM). It demonstrates that both approaches achieve competitive or superior performance with only a fraction of the training data, relative to full-data baselines, while caching LLM-derived features can align online latency with CRM backbones. The work provides a principled, two-pronged path toward practical, sample-efficient LLM-enhanced RS and offers insights into latency handling and backbone compatibility. It also points to future directions in few-shot sample selection and broader applications such as code snippet recommendations.
Abstract
Large language models (LLMs) have achieved remarkable progress in the field of natural language processing (NLP), demonstrating remarkable abilities in producing text that resembles human language for various tasks. This opens up new opportunities for employing them in recommender systems (RSs). In this paper, we specifically examine the sample efficiency of LLM-enhanced recommender systems, which pertains to the model's capacity to attain superior performance with a limited quantity of training data. Conventional recommendation models (CRMs) often need a large amount of training data because of the sparsity of features and interactions. Hence, we propose and verify our core viewpoint: Large Language Models Make Sample-Efficient Recommender Systems. We propose a simple yet effective framework (i.e., Laser) to validate the viewpoint from two aspects: (1) LLMs themselves are sample-efficient recommenders; and (2) LLMs, as feature generators and encoders, make CRMs more sample-efficient. Extensive experiments on two public datasets show that Laser requires only a small fraction of training samples to match or even surpass CRMs that are trained on the entire training set, demonstrating superior sample efficiency.