SurveyAgent-HKA: A multi-agent framework for scientific survey generation with LLMs and human knowledge augmentation

arXiv 2026 52.7 method

TLDR

SurveyAgent-HKA is a multi-agent framework for scientific survey generation that incorporates human-written survey outlines and peer-review comments to improve reference coverage and content quality.

Reasoning

The paper presents a clear multi-agent framework addressing limitations of single-source retrieval and one-pass LLM generation, with experiments showing gains in citation quality, structural consistency, and content quality. However, evaluation is limited to two domains and the abstract does not provide detailed quantitative results or comparisons of the peer-review revision component.

Read-first score

Read-first score 52.7, weighted from topical fit, citation, graph, method, reproducibility, and recency signals. Original total remains 63.

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=baseline,experiment

Topical relevance 42%

52.5

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=none

Field roles

Frontier

Rank sensitivity

Stability: volatile; rank range: 91.

Keyword Scores

survey generation

literature review agent

paper writing agent

research automation

automated research

AI for scientific research

autonomous research agent

AI scientist

automated scientific discovery

scientific discovery agent

automated experimentation

experiment design agent

Deep Analysis

Innovations

Proposes SurveyAgent-HKA, a multi-agent framework that decomposes scientific survey generation into sub-tasks handled by LLM-powered agents.
Incorporates human knowledge augmentation from published surveys and peer-review comments, rather than relying only on retrieved literature.
Uses multi-source paper retrieval and topic clustering to construct an initial outline.
Refines the initial outline using outlines from related human-written surveys.
Retrieves and re-ranks topic-focused papers based on the refined outline to support grounded drafting.
Identifies common expert issues from peer-review comments of published surveys to guide survey revisions.

Methodology

SurveyAgent-HKA is a multi-agent framework that decomposes scientific survey generation into subtasks handled by LLM-powered agents. It retrieves relevant papers from multiple sources, clusters them to identify key topics and construct an initial outline, refines that outline using related human-written survey outlines, then retrieves and re-ranks topic-focused papers for drafting. It also uses common issues found in peer-review comments from published surveys to guide revisions, and is evaluated on two domains against mainstream baselines.

Key Results

In experiments on two domains, SurveyAgent-HKA outperforms mainstream baselines in citation quality, structural consistency, and content quality, while also being efficient in time and cost.

Resources

Paper