# LLM-Rec: Personalized Recommendation via Prompting Large Language Models

**Authors:** Hanjia Lyu, Song Jiang, Hanqing Zeng, Yinglong Xia, Jiebo Luo  
**arXiv:** [2307.15780](https://arxiv.org/abs/2307.15780)  
**Date:** 2023-07-24  
**Summary page:** https://www.summarizepaper.com/en/paper/2307.15780v1/

## Key points

- Authors investigate prompting strategies to enhance personalized content recommendation performance using large language models (LLMs)
- Proposed approach called LLM-Rec incorporates four distinct prompting strategies: basic prompting, recommendation-driven prompting, engagement-guided prompting, and recommendation-driven + engagement-guided prompting
- Empirical experiments show that combining original content description with augmented input text generated by LLM using these prompting strategies significantly improves recommendation performance
- Importance of incorporating diverse prompts and input augmentation techniques for enhancing capabilities of large language models in personalized content recommendation emphasized
- LLM-Rec offers a comprehensive framework for improving recommendation systems by considering recommendations and user engagement
- Findings have implications for e-commerce platforms, news aggregators, and social media platforms relying on personalized content recommendations
- Research contributes to advancing the field of personalized content recommendation by highlighting potential of diverse prompts and input augmentation techniques with large language models
- Proposed LLM-Rec approach presents a promising avenue for future research in enhancing recommendation systems' performance and tailoring them to individual users' preferences.

## Summary

In the study titled "LLM-Rec: Personalized Recommendation via Prompting Large Language Models," authors Hanjia Lyu, Song Jiang, Hanqing Zeng, Yinglong Xia, and Jiebo Luo investigate various prompting strategies to enhance personalized content recommendation performance using large language models (LLMs). They propose a novel approach called LLM-Rec, which incorporates four distinct prompting strategies: basic prompting, recommendation-driven prompting, engagement-guided prompting, and recommendation-driven + engagement-guided prompting. Through empirical experiments, the authors demonstrate that combining the original content description with augmented input text generated by LLM using these prompting strategies significantly improves recommendation performance. This finding emphasizes the importance of incorporating diverse prompts and input augmentation techniques to enhance the capabilities of large language models for personalized content recommendation. The study provides valuable insights into how prompt engineering can improve the effectiveness of LLMs in recommending personalized content. By considering different aspects such as recommendations and user engagement, LLM-Rec offers a comprehensive framework for improving recommendation systems. The findings have implications for various applications that rely on personalized content recommendations including e-commerce platforms, news aggregators and social media platforms. Overall, this research contributes to advancing the field of personalized content recommendation by highlighting the potential of incorporating diverse prompts and input augmentation techniques with large language models. The proposed LLM-Rec approach presents a promising avenue for future research in enhancing recommendation systems' performance and tailoring them to individual users' preferences.