Reproduce “Deep Academic Survey: Stateful Agentic Closed-Loop Paradigm for Academic Survey Automation” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Zhikai Xu, Zhucun Xue, Teng Hu, Yabiao Wang, Yong Liu, Jiangning Zhang
Researchers from Zhejiang University and Shanghai Jiao Tong University introduce Deep Academic Survey (DAS), a stateful agentic closed-loop framework that automates the generation of publication-oriented academic surveys by integrating literature organization, hierarchical drafting, and robust validation. DAS achieved an overall score of 4.34 on a new 30-topic benchmark, outperforming other automated systems.
Key insights
- A stateful agentic framework with an explicit, shared manuscript state enables more consistent and reliable survey construction by allowing targeted review feedback to reactivate only affected writing states.
- Utilizing rich, pre-computed, survey-oriented paper representations (DAS-2M) is crucial for deeper paper understanding and supports more sophisticated, fine-grained literature organization and drafting than shallow summaries.
- The combination of a scoped semantic review-and-repair loop and deterministic validation is essential for ensuring both the semantic coherence and technical compilability of publication-oriented academic surveys.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.