Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

AutoSurvey2: Empowering Researchers with Next Level Automated Literature Surveys

Siyi Wu, Chiaxin Liang, Ziqian Bi, Leyi Zhao, Tianyang Wang, Junhao Song, Yichao Zhang, Keyu Chen, Benji Peng, Xinyuan Song

TL;DR

AutoSurvey2 addresses the challenge of producing up-to-date, well-structured survey papers in fast-moving AI/ML fields by introducing a four-stage, retrieval-augmented pipeline with a graph-based execution model and multi-LLM evaluation. The system combines semantic retrieval, modular generation, and automated post-processing to produce publication-ready IEEE-formatted LaTeX documents, while maintaining reproducibility through a shared global state. A Judge Agent-based evaluation framework assesses Coverage, Structure, and Relevance, demonstrating improvements over baselines. Experimental results show AutoSurvey2 achieving higher structural coherence, topical relevance, and citation fidelity, establishing a scalable foundation for automated scholarly writing with potential for broad impact in research synthesis.

Abstract

The rapid growth of research literature, particularly in large language models (LLMs), has made producing comprehensive and current survey papers increasingly difficult. This paper introduces autosurvey2, a multi-stage pipeline that automates survey generation through retrieval-augmented synthesis and structured evaluation. The system integrates parallel section generation, iterative refinement, and real-time retrieval of recent publications to ensure both topical completeness and factual accuracy. Quality is assessed using a multi-LLM evaluation framework that measures coverage, structure, and relevance in alignment with expert review standards. Experimental results demonstrate that autosurvey2 consistently outperforms existing retrieval-based and automated baselines, achieving higher scores in structural coherence and topical relevance while maintaining strong citation fidelity. By combining retrieval, reasoning, and automated evaluation into a unified framework, autosurvey2 provides a scalable and reproducible solution for generating long-form academic surveys and contributes a solid foundation for future research on automated scholarly writing. All code and resources are available at https://github.com/annihi1ation/auto_research.

AutoSurvey2: Empowering Researchers with Next Level Automated Literature Surveys

TL;DR

AutoSurvey2 addresses the challenge of producing up-to-date, well-structured survey papers in fast-moving AI/ML fields by introducing a four-stage, retrieval-augmented pipeline with a graph-based execution model and multi-LLM evaluation. The system combines semantic retrieval, modular generation, and automated post-processing to produce publication-ready IEEE-formatted LaTeX documents, while maintaining reproducibility through a shared global state. A Judge Agent-based evaluation framework assesses Coverage, Structure, and Relevance, demonstrating improvements over baselines. Experimental results show AutoSurvey2 achieving higher structural coherence, topical relevance, and citation fidelity, establishing a scalable foundation for automated scholarly writing with potential for broad impact in research synthesis.

Abstract

The rapid growth of research literature, particularly in large language models (LLMs), has made producing comprehensive and current survey papers increasingly difficult. This paper introduces autosurvey2, a multi-stage pipeline that automates survey generation through retrieval-augmented synthesis and structured evaluation. The system integrates parallel section generation, iterative refinement, and real-time retrieval of recent publications to ensure both topical completeness and factual accuracy. Quality is assessed using a multi-LLM evaluation framework that measures coverage, structure, and relevance in alignment with expert review standards. Experimental results demonstrate that autosurvey2 consistently outperforms existing retrieval-based and automated baselines, achieving higher scores in structural coherence and topical relevance while maintaining strong citation fidelity. By combining retrieval, reasoning, and automated evaluation into a unified framework, autosurvey2 provides a scalable and reproducible solution for generating long-form academic surveys and contributes a solid foundation for future research on automated scholarly writing. All code and resources are available at https://github.com/annihi1ation/auto_research.

Paper Structure

This paper contains 15 sections, 19 equations, 1 figure, 2 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. Methodology
  4. Overview
  5. Database Construction
  6. Stage 1: Research Planning
  7. Stage 2: Research Phase
  8. Stage 3: Generation Phase
  9. Stage 4: Post-processing
  10. Prompt Design
  11. Experiments
  12. Results
  13. Conclusion
  14. Limitations
  15. Ethical Statement

Figures (1)

  • Figure 1: Overall workflow of AutoSurvey2. The system operates through four sequential stages organized as a graph-based state machine: Database Construction (prerequisite), Research Planning (Stage 1), Research Phase (Stage 2), Generation Phase (Stage 3), and Post-processing (Stage 4). Each stage comprises specialized functional nodes that communicate through a shared global state $\Sigma$.