Insightrag uses author citations to improve scientific survey writing
InsightRAG: Aggregating Distributed Scholarly Judgments for Scientific Survey Generation
Artificial IntelligenceDigital Libraries
Summary
Writing overviews of scientific research is hard because there are so many papers and ideas to organize. The authors of this paper noticed that researchers’ own descriptions of related work show how they understand and link different studies. They built InsightRAG, a system that follows these human-made citation paths to collect and organize relevant papers better. This helps make clearer overviews with reliable references and meaningful categories based on how researchers connect their work. Tests on ten topics showed this approach improved the scope and structure of automatically generated surveys.
What this means in practice
- •For scientific publishing teams: Generate more accurate and organized survey sections by using author citation patterns to guide literature selection and summary structuring.
- •For academic software developers: Create tools that assist researchers in writing literature reviews by automatically aggregating related work based on citation relations given by authors.
Authors
Yuquan Lin, Wenshuo Zhang, Yilin Ye, Shuchang Xu, Zhao Chen, Weiqi Wang, Huamin Qu
Abstract
The rapid growth of scientific literature makes survey writing increasingly challenging, motivating automatic survey generation. Yet, existing systems must search an open space of papers and an open space of concepts with no trace of how the community itself has navigated them, leading to incomplete scope, weak taxonomies, and unreliable citations. Our core insight is that author-written Related Work sections contain the human judgments needed to address both limitations: authors curate the citations relevant to their contributions and explain how the cited works relate, yielding paths for scope reconstruction and relational evidence for organization. We propose InsightRAG. Its insight-guided retrieval follows author-supported citation relations to reconstruct a verified research scope and structure the resulting relational evidence. It aggregates these judgments under a minimum-conflict objective into a taxonomy that reflects recurring author-written perspectives in the retrieved literature, guides generation, and bounds each section's evidence. Results on ten topics show gains in scope reconstruction, taxonomy quality, citation support, and reference reliability.