Awesome Auto Research Hub Papers · Datasets · Projects
← Back to papers

SciAtlas: A Large-Scale Knowledge Graph for Automated Scientific Research

arXiv 2026 43.8 method

TLDR

SciAtlas is a large-scale multi-disciplinary knowledge graph with 43M papers and a neuro-symbolic retrieval algorithm to support automated scientific research.

Reasoning

The paper presents a substantial knowledge graph and retrieval method, but the abstract lacks concrete experimental results or comparisons, and the claimed applications are only outlined as directions.

Read-first score

Read-first score 43.8, weighted from topical fit, citation, graph, method, reproducibility, and recency signals. Original total remains 46.

Recency 8%
100

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

Methodology quality 25%
40

Screens visible abstract and analysis fields for experiment, dataset, baseline, metric, and limitation evidence. markers=none

Topical relevance 42%
38.3

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%
38

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

Field roles

FrontierBridge

Rank sensitivity

Stability: volatile; rank range: 62.

Keyword Scores

AI for scientific research
8
automated research
7
research automation
6
literature review agent
5
automated scientific discovery
4
survey generation
4
scientific discovery agent
4
autonomous research agent
3
AI scientist
2
automated experimentation
1
experiment design agent
1
paper writing agent
1

Deep Analysis

Innovations

  • Large-scale multi-disciplinary heterogeneous knowledge graph (SciAtlas) integrating 43M papers, 157M entities, and 3B triplets
  • Neuro-symbolic retrieval algorithm with tri-path collaborative recall and graph reranking for deterministic association discovery
  • Application of SciAtlas as a cognitive map for automated scientific research tasks including literature review, trend synthesis, idea positioning, and trajectory exploration

Methodology

SciAtlas is built as a heterogeneous knowledge graph from over 43 million papers across 26 disciplines, yielding 157 million entities and 3 billion triplets. A neuro-symbolic retrieval algorithm combining tri-path collaborative recall and graph reranking is developed to transition from semantic matching to deterministic association discovery. The system is applied to literature review, automated research trend synthesis, idea positioning, and academic trajectory exploration.

Key Results

The knowledge graph construction achieves a scale of 43M papers, 157M entities, and 3B triplets; the retrieval algorithm enables deterministic association discovery, and application demonstrations suggest reduced reasoning costs for automated research.

Tags

AICLIRLG