DeepSurvey: Enhancing Analytical Depth and Citation Reliability in Automated Survey Generation

arXiv 2026 66.2 method

TLDR

DeepSurvey is an agentic system for automated survey generation that improves analytical depth and citation reliability using full-text analysis, cross-paper modeling, and evidence-constrained citation.

Reasoning

The paper presents a well-motivated approach addressing key limitations in automated survey generation, with strong empirical results including expert preference over human-written surveys. However, its focus is narrow (survey generation) and may not generalize to broader scientific discovery tasks without full-text access.

Read-first score

Read-first score 66.2, weighted from topical fit, citation, graph, method, reproducibility, and recency signals. Original total remains 83.

Recency 8%

100

Uses a gentle age decay so recent papers surface without erasing older foundations. 2026

Methodology quality 25%

Screens visible abstract and analysis fields for experiment, dataset, baseline, metric, and limitation evidence. markers=analysis,baseline,experiment

Topical relevance 42%

69.2

Uses existing LLM keyword relevance scores normalized to 0-100. AI scientist,automated scientific discovery,autonomous research agent,automated research,literature review agent,survey generation,automated experimentation,experiment design agent,AI for scientific research,paper writing agent,research automation,scientific discovery agent

Reproducibility 25%

Screens links and visible text for paper, code, dataset, artifact, and repository signals. pdf=True; code=False; dataset=False; markers=code,repository

Field roles

FrontierMethodology anchor

Rank sensitivity

Stability: volatile; rank range: 48.

Keyword Scores

survey generation

literature review agent

paper writing agent

autonomous research agent

AI for scientific research

research automation

AI scientist

automated research

scientific discovery agent

automated scientific discovery

automated experimentation

experiment design agent

Deep Analysis

Innovations

Extracting structured keynotes from full-text papers to enhance analytical depth
Modeling cross-paper relationships through clustering and comparative analysis
Integrating code-repository analysis to recover implementation-level details
Combining citation-graph expansion with hybrid filtering for topic-focussed retrieval
Enforcing evidence-constrained citation assignment
Deploying multi-granularity agentic refinement to validate citation-claim alignment

Methodology

DeepSurvey is an agentic system that enhances depth by extracting structured keynotes from full-text papers, modeling cross-paper relationships via clustering and comparative analysis, and integrating code-repository analysis. For citation reliability, it uses citation-graph expansion with hybrid filtering, evidence-constrained citation assignment, and multi-granularity agentic refinement to validate citation-claim alignment.

Key Results

DeepSurvey achieves the highest content score (8.644/10), citation quality gains of 12.3% recall and 9.3% precision over the strongest baseline, robust cross-domain generalization (0.14 vs 0.22 to 0.69 CS-to-non-CS drop), and is preferred over human-written surveys by domain experts (83.3% overall quality, 100% content depth).

Resources

Paper