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

ResearchStudio-Idea: An Evidence-Grounded Research-Ideation Skill Suite from ML Conference Outcomes

arXiv 2026 46.9 method, system, application

TLDR

A skill suite for evidence-grounded research ideation using patterns derived from ML conference papers.

Reasoning

The paper introduces a structured approach to research ideation with real conference data, which is a strength. However, it focuses only on the ideation phase and lacks full automation or experimental validation, limiting its scope.

Read-first score

Read-first score 46.9, weighted from topical fit, citation, graph, method, reproducibility, and recency signals. Original total remains 39.

Recency 8%
100

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

Methodology quality 25%
70

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

Topical relevance 42%
32.5

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%
30

Screens links and visible text for paper, code, dataset, artifact, and repository signals. pdf=True; code=False; dataset=False; markers=none

Field roles

FrontierMethodology anchor

Rank sensitivity

Stability: volatile; rank range: 27.

Keyword Scores

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

Deep Analysis

Innovations

  • ResearchStudio-Idea skill suite with Paper-Search (multi-source literature search), Scoop-Check (prior-art collision checker), and IdeaSpark (end-to-end ideation skill)
  • IdeaSpark workflow that composes evidence grounding, pattern-guided generation, collision retrieval, audit, and idea-card rendering into a traceable research proposal
  • Derivation of 15 reusable ideation patterns from analysis of 1,947 ML conference papers (including Oral, high-citation, and rejected submissions) across ICLR, ICML, NeurIPS 2021–2025
  • Operationalization of each ideation pattern as a structured card containing research contexts, bottleneck types, differentiation strategies, supporting precedents, and common failure modes

Methodology

The authors collected a corpus of 1,947 machine learning conference papers (ICLR, ICML, NeurIPS 2021–2025) including Oral, high-citation, and rejected submissions, and analyzed outcomes to extract 31 ideation sub-patterns consolidated into 15 reusable patterns. IdeaSpark takes a research problem and evidence bundle, evaluates evidence readiness, reconstructs context, identifies bottlenecks, selects patterns, instantiates a candidate direction, retrieves conflicting prior work, and performs outcome-informed auditing. Evaluation uses blind automated-judge comparisons against no-skill and generic-skill baselines.

Key Results

IdeaSpark consistently produces stronger research proposals than no-skill and generic-skill baselines while maintaining competitive novelty.

Resources

Paper

Tags