How does Scientific Taste differ from standard LLM ideation?

While standard LLMs often confuse unfamiliarity with novelty, this skill uses a 7-dimension framework and retrieval-grounded audits to focus on structural impact and real-world feasibility.

What is the benefit of a 'novelty audit'?

The audit command systematically retrieves the closest existing papers to your idea and forces a comparison to ensure there is a clear, articulable difference, preventing accidental duplication.

Can this skill help me prepare for specific conferences?

Yes, it includes specific modules for venues like NeurIPS, ICML, CHI, and UIST to ensure your idea aligns with specific community standards and trajectories.

What is 'Dual-Path' generation?

It is a methodology that generates ideas starting from both real-world problems (bottlenecks) and unique technical insights (surprising findings), leading to better-balanced research topics.

Scientific Taste Research Critic

Name: Scientific Taste Research Critic
Author: Carolzhangzz

byCarolzhangzz

0•

데이터 과학 및 ML

Evaluates and generates high-impact research ideas using a 7-dimension framework based on senior-reviewer intuition and academic literature.

Scientific Taste is a specialized Claude Code skill designed to help researchers distinguish paradigm-shifting ideas from incremental variants. By integrating methodologies from over 16 papers on automated research ideation, it provides senior-reviewer-level feedback on novelty, feasibility, and trajectory alignment. The skill helps users move beyond simple 'checklist scoring' by performing literature-grounded audits, position-swap debiased comparisons, and dual-path generation to ensure research concepts are both novel and practically impactful for top-tier venues like NeurIPS, ICML, CHI, and UIST.

주요 기능

01Position-swap debiased pairwise comparisons between research ideas

027-Dimension Taste Evaluation for deep structural impact analysis

03Dual-path generation pipeline (Problem-First vs. Insight-First)

040 GitHub stars

05Literature-grounded novelty auditing to prevent 'reinventing the wheel'

06Venue-specific criteria modules for major conferences like NeurIPS and CHI

사용 사례

01Generating novel research directions from specific real-world bottlenecks

02Stress-testing a research concept before committing months of work

03Comparing two competing research directions to identify the highest impact path

주요 기능

01Position-swap debiased pairwise comparisons between research ideas

027-Dimension Taste Evaluation for deep structural impact analysis

03Dual-path generation pipeline (Problem-First vs. Insight-First)

040 GitHub stars

05Literature-grounded novelty auditing to prevent 'reinventing the wheel'

06Venue-specific criteria modules for major conferences like NeurIPS and CHI

사용 사례

01Generating novel research directions from specific real-world bottlenecks

02Stress-testing a research concept before committing months of work

03Comparing two competing research directions to identify the highest impact path