Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Information Retrieval Meets Large Language Models: A Strategic Report from Chinese IR Community

Qingyao Ai, Ting Bai, Zhao Cao, Yi Chang, Jiawei Chen, Zhumin Chen, Zhiyong Cheng, Shoubin Dong, Zhicheng Dou, Fuli Feng, Shen Gao, Jiafeng Guo, Xiangnan He, Yanyan Lan, Chenliang Li, Yiqun Liu, Ziyu Lyu, Weizhi Ma, Jun Ma, Zhaochun Ren, Pengjie Ren, Zhiqiang Wang, Mingwen Wang, Ji-Rong Wen, Le Wu, Xin Xin, Jun Xu, Dawei Yin, Peng Zhang, Fan Zhang, Weinan Zhang, Min Zhang, Xiaofei Zhu

TL;DR

This paper surveys how Large Language Models can transform Information Retrieval by rethinking IR core values and proposing a triadic paradigm that unites Humans, IR systems, and LLMs. It details how LLMs can enhance IR components (user modeling, indexing, matching, evaluation, interaction) while IR can enhance LLMs via retrieval-augmented pretraining, adapters, and grounding for black-box models. The authors present a unified framework emphasizing three modules and discuss practical challenges—computational cost, credibility, domain adaptation, and ethics—along with future research directions. The work provides a strategic roadmap for integrating LLMs into IR workflows to support more intelligent, trustworthy, and user-centered information services.

Abstract

The research field of Information Retrieval (IR) has evolved significantly, expanding beyond traditional search to meet diverse user information needs. Recently, Large Language Models (LLMs) have demonstrated exceptional capabilities in text understanding, generation, and knowledge inference, opening up exciting avenues for IR research. LLMs not only facilitate generative retrieval but also offer improved solutions for user understanding, model evaluation, and user-system interactions. More importantly, the synergistic relationship among IR models, LLMs, and humans forms a new technical paradigm that is more powerful for information seeking. IR models provide real-time and relevant information, LLMs contribute internal knowledge, and humans play a central role of demanders and evaluators to the reliability of information services. Nevertheless, significant challenges exist, including computational costs, credibility concerns, domain-specific limitations, and ethical considerations. To thoroughly discuss the transformative impact of LLMs on IR research, the Chinese IR community conducted a strategic workshop in April 2023, yielding valuable insights. This paper provides a summary of the workshop's outcomes, including the rethinking of IR's core values, the mutual enhancement of LLMs and IR, the proposal of a novel IR technical paradigm, and open challenges.

Information Retrieval Meets Large Language Models: A Strategic Report from Chinese IR Community

TL;DR

This paper surveys how Large Language Models can transform Information Retrieval by rethinking IR core values and proposing a triadic paradigm that unites Humans, IR systems, and LLMs. It details how LLMs can enhance IR components (user modeling, indexing, matching, evaluation, interaction) while IR can enhance LLMs via retrieval-augmented pretraining, adapters, and grounding for black-box models. The authors present a unified framework emphasizing three modules and discuss practical challenges—computational cost, credibility, domain adaptation, and ethics—along with future research directions. The work provides a strategic roadmap for integrating LLMs into IR workflows to support more intelligent, trustworthy, and user-centered information services.

Abstract

The research field of Information Retrieval (IR) has evolved significantly, expanding beyond traditional search to meet diverse user information needs. Recently, Large Language Models (LLMs) have demonstrated exceptional capabilities in text understanding, generation, and knowledge inference, opening up exciting avenues for IR research. LLMs not only facilitate generative retrieval but also offer improved solutions for user understanding, model evaluation, and user-system interactions. More importantly, the synergistic relationship among IR models, LLMs, and humans forms a new technical paradigm that is more powerful for information seeking. IR models provide real-time and relevant information, LLMs contribute internal knowledge, and humans play a central role of demanders and evaluators to the reliability of information services. Nevertheless, significant challenges exist, including computational costs, credibility concerns, domain-specific limitations, and ethical considerations. To thoroughly discuss the transformative impact of LLMs on IR research, the Chinese IR community conducted a strategic workshop in April 2023, yielding valuable insights. This paper provides a summary of the workshop's outcomes, including the rethinking of IR's core values, the mutual enhancement of LLMs and IR, the proposal of a novel IR technical paradigm, and open challenges.
Paper Structure (28 sections, 1 figure)

This paper contains 28 sections, 1 figure.

Table of Contents

  1. Introduction
  2. Rethinking Core Values of IR
  3. Scientific Connotation of IR
  4. User Perspective
  5. Ranking Perspective
  6. Corpus Perspective
  7. Boundaries and Extensions of IR
  8. Ranking or Generation?
  9. For Human or For Machine?
  10. For Finding or For Decision?
  11. Going beyond Cyberspace?
  12. Large Language Models for IR
  13. User Modeling
  14. Indexing
  15. Matching/Ranking
  16. ...and 13 more sections

Figures (1)

  • Figure 1: Our proposed new IR technical paradigm introduces three key elements: Humans, IR models, and LLMs. The synergistic relationship among them not only facilitates mutual enhancement but also enables the fulfillment of new tasks as an organic whole.