Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Large Language Models for Generative Information Extraction: A Survey

Derong Xu, Wei Chen, Wenjun Peng, Chao Zhang, Tong Xu, Xiangyu Zhao, Xian Wu, Yefeng Zheng, Yang Wang, Enhong Chen

TL;DR

This survey analyzes the rapid emergence of generative information extraction (IE) using large language models (LLMs). It introduces dual taxonomies—IE subtasks (NER/RE/EE/Unified IE) and LLM-centric techniques (data augmentation, prompting, fine-tuning, decoding constraints)—and contrasts natural-language versus code-based universal IE frameworks. Empirically, universal and prompt-aware approaches often outperform task-specific discriminative models, especially in complex or cross-domain scenarios, while data-efficient strategies remain critical for low-resource domains. The work also surveys domain-specific applications, evaluation insights, and directions toward scalable, robust, and Open IE-enabled IE with LLMs.

Abstract

Information extraction (IE) aims to extract structural knowledge from plain natural language texts. Recently, generative Large Language Models (LLMs) have demonstrated remarkable capabilities in text understanding and generation. As a result, numerous works have been proposed to integrate LLMs for IE tasks based on a generative paradigm. To conduct a comprehensive systematic review and exploration of LLM efforts for IE tasks, in this study, we survey the most recent advancements in this field. We first present an extensive overview by categorizing these works in terms of various IE subtasks and techniques, and then we empirically analyze the most advanced methods and discover the emerging trend of IE tasks with LLMs. Based on a thorough review conducted, we identify several insights in technique and promising research directions that deserve further exploration in future studies. We maintain a public repository and consistently update related works and resources on GitHub (\href{https://github.com/quqxui/Awesome-LLM4IE-Papers}{LLM4IE repository})

Large Language Models for Generative Information Extraction: A Survey

TL;DR

This survey analyzes the rapid emergence of generative information extraction (IE) using large language models (LLMs). It introduces dual taxonomies—IE subtasks (NER/RE/EE/Unified IE) and LLM-centric techniques (data augmentation, prompting, fine-tuning, decoding constraints)—and contrasts natural-language versus code-based universal IE frameworks. Empirically, universal and prompt-aware approaches often outperform task-specific discriminative models, especially in complex or cross-domain scenarios, while data-efficient strategies remain critical for low-resource domains. The work also surveys domain-specific applications, evaluation insights, and directions toward scalable, robust, and Open IE-enabled IE with LLMs.

Abstract

Information extraction (IE) aims to extract structural knowledge from plain natural language texts. Recently, generative Large Language Models (LLMs) have demonstrated remarkable capabilities in text understanding and generation. As a result, numerous works have been proposed to integrate LLMs for IE tasks based on a generative paradigm. To conduct a comprehensive systematic review and exploration of LLM efforts for IE tasks, in this study, we survey the most recent advancements in this field. We first present an extensive overview by categorizing these works in terms of various IE subtasks and techniques, and then we empirically analyze the most advanced methods and discover the emerging trend of IE tasks with LLMs. Based on a thorough review conducted, we identify several insights in technique and promising research directions that deserve further exploration in future studies. We maintain a public repository and consistently update related works and resources on GitHub (\href{https://github.com/quqxui/Awesome-LLM4IE-Papers}{LLM4IE repository})
Paper Structure (24 sections, 3 equations, 5 figures, 8 tables)

This paper contains 24 sections, 3 equations, 5 figures, 8 tables.

Table of Contents

  1. Introduction
  2. Preliminaries of Generative IE
  3. LLMs for Different Information Extraction Tasks
  4. Named Entity Recognition
  5. Relation Extraction
  6. Event Extraction
  7. Universal Information Extraction
  8. Summaries of Tasks
  9. Techniques of LLMs for Generative IE
  10. Data Augmentation
  11. Prompt Design
  12. Zero-shot Learning
  13. Constrained Decoding Generation
  14. Few-shot Learning
  15. Supervised Fine-tuning
  16. ...and 9 more sections

Figures (5)

  • Figure 1: LLMs have been extensively explored for generative IE. These studies encompass various IE techniques, specialized frameworks designed for a single subtask, and universal frameworks capable of addressing multiple subtasks simultaneously.
  • Figure 2: Taxonomy of research in generative IE using LLMs. Some papers have been omitted due to space limitations.
  • Figure 3: Examples of different IE tasks.
  • Figure 4: The comparison of prompts of NL-LLMs and Code-LLMs for universal IE. Both NL-based and code-based methods attempt to construct a universal schema, but they differ in terms of prompt format and the way they utilize the generation capabilities of LLMs. This figure is adopted from instructuie and code4uie.
  • Figure 5: Comparison of data augmentation methods.