Agentic AutoSurvey: Let Agentic LLM Survey LLMs

Agents4Science 2025 Conference Submission215 Authors

Keywords: multi-agent systems, large language models, automated survey generation, literature review, paper search, topic mining and clustering

TL;DR: A four-agent LLM pipeline searches, clusters, writes, and evaluates to automatically produce high-quality literature surveys with broad citation coverage and clear synthesis.

Abstract: The exponential growth of scientific literature poses unprecedented challenges for researchers attempting to synthesise knowledge across rapidly evolving fields. We present \textbf{Agentic AutoSurvey}, a multi-agent framework for automated survey generation that addresses fundamental limitations in existing approaches. Our system employs four specialised agents (Paper Search Specialist, Topic Mining \& Clustering, Academic Survey Writer, and Quality Evaluator) working in concert to generate comprehensive literature surveys with superior synthesis quality. Through experiments on six representative LLM research topics from COLM 2024 categories, we demonstrate that our multi-agent approach achieves significant improvements over existing baselines, scoring 8.18/10 compared to AutoSurvey's 4.77/10. The multi-agent architecture enables processing of large paper collections (up to 847 papers) while maintaining high citation coverage (80\%+) and synthesis quality through specialized agent orchestration. Our comprehensive 12-dimensional evaluation framework provides nuanced quality assessment beyond traditional metrics, revealing that specialized agent decomposition produces surveys with superior organization, synthesis integration, and critical analysis compared to existing automated approaches. These findings demonstrate that multi-agent architectures represent a meaningful advancement for automated literature survey generation in rapidly evolving scientific domains.

Submission Number: 215