# A Survey of Generative Search and Recommendation in the Era of Large Language Models

**Authors:** Yongqi Li, Xinyu Lin, Wenjie Wang, Fuli Feng, Liang Pang, Wenjie Li, Liqiang Nie, Xiangnan He, Tat-Seng Chua  
**arXiv:** [2404.16924](https://arxiv.org/abs/2404.16924)  
**Date:** 2024-04-25  
**Summary page:** https://www.summarizepaper.com/en/paper/2404.16924v1/

## Key points

- Search and recommendation systems are crucial for meeting users' information needs in the era of large language models.
- Technological advancements in search and recommendation systems have been significant, especially with the adoption of machine learning and deep learning paradigms.
- The emergence of generative search and recommendation, driven by superintelligent generative large language models, offers a dynamic approach to addressing the matching problem.
- Authors Yongqi Li, Xinyu Lin, Wenjie Wang, Fuli Feng, Liang Pang, Wenjie Li, Liqiang Nie, Xiangnan He,and Tat-Seng Chua provide a comprehensive overview of generative search and recommendation systems in their survey paper.
- The authors present a unified framework for understanding these systems instead of categorizing existing works.
- Different stages within the framework highlight strengths and weaknesses of generative search and recommendation systems.
- Unique challenges between generative search and recommendation processes are addressed in the paper.
- Open research problems and future directions for advancing generative search and recommendation systems are identified by the authors.
- Leveraging large language models has the potential to enhance search and recommendation capabilities on the web.

## Summary

The importance of search and recommendation systems in meeting users' information needs cannot be overstated in the era of large language models. These two fundamental infrastructures are interconnected, both aiming to effectively match queries with documents or users with items. Over the past few decades, significant technological advancements have been made in search and recommendation systems, particularly with the rise of machine learning and deep learning paradigms. Recently, a new paradigm known as generative search and recommendation has emerged, driven by superintelligent generative large language models. This innovative approach seeks to address the matching problem in a more dynamic and generative manner. In their comprehensive survey paper, authors Yongqi Li, Xinyu Lin, Wenjie Wang, Fuli Feng, Liang Pang, Wenjie Li, Liqiang Nie, Xiangnan He,and Tat-Seng Chua provide an insightful overview of this emerging paradigm within information systems. They present a unified framework for understanding generative search and recommendation systems instead of simply categorizing existing works in this field. The authors break down these systems into different stages within the framework to highlight their respective strengths and weaknesses. Additionally,the paper distinguishes between generative search and recommendation processes by addressing their unique challenges. Furthermore,the authors identify open research problems and future directions for advancing generative search and recommendation systems.By envisioning the next evolution in information-seeking paradigms they pave the way for further innovation in this critical area of information retrieval technology.The collaborative effort of these researchers sheds light on the potential of leveraging large language models for enhancing searchandrecommendation capabilities on the web.