# Stem Literature Review

> Literature review and research-gap analysis for STEM graduate papers, theses, and journal or conference manuscripts. Use when users need to search, organize, compare, synthesize, or write related work; identify research gaps; build a citation matrix; position a contribution against IEEE, ACM, Elsevier, Springer, Nature-family, arXiv, or field-specific literature. Trigger phrases include literature review, related work, research gap, novelty, contribution positioning, citation matrix, survey, state of the art, and Chinese requests about wenxian zongshu or yanjiu kongbai.

- Skill: `xiangdong-415/stem-literature-review` (Agent Skill, multi-file: 2 files)
- Install (CLI): `npx skillmds@latest add xiangdong-415/stem-literature-review`
- Raw SKILL.md: https://api.skillmd.com/api/skills/xiangdong-415/stem-literature-review/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Marketing & Growth
- Author: Xiangdong-415 (https://skillmd.com/u/xiangdong-415)
- Updated: 2026-09-22
- Page: https://skillmd.com/skills/xiangdong-415/stem-literature-review

---


# STEM Literature Review

## Purpose

Use this skill to turn a STEM research idea, draft, or paper list into a defensible literature review and contribution position. It is designed for engineering, computer science, materials, biomedical, control, communication, energy, chemistry, physics, and other technical graduate work.

## Inputs To Ask For

If missing, ask only for the minimum needed:

- Research topic or working title.
- Field and target venue type, such as IEEE journal, ACM conference, Elsevier journal, thesis, or arXiv.
- Known keywords, seed papers, or competing methods.
- User's actual contribution, method, dataset, model, experiment, or result.

## Workflow

1. Clarify the research object.
   - Extract core problem, application domain, technical method, and evaluation metric.
   - Separate "what problem is solved" from "what method is used".

2. Build search strategy.
   - Generate keyword groups for problem, method, domain, metrics, and synonyms.
   - Include recent work from the last 3-5 years and foundational work.
   - Prefer official sources and primary literature when browsing is available: publisher pages, arXiv, PubMed, IEEE Xplore, ACM DL, ScienceDirect, SpringerLink, official project pages, and Google Scholar-style metadata.

3. Create a literature matrix.
   - Columns: citation, problem, method, data or experiment, metrics, main result, limitation, relation to user work.
   - Group papers by technical approach, not only by year.
   - Mark direct competitors separately from background papers.

4. Identify gaps.
   - Evidence gap: no measurement or no data in an important setting.
   - Method gap: existing methods fail under a constraint such as scale, latency, robustness, interpretability, cost, or generalization.
   - Evaluation gap: weak baselines, missing ablation, narrow benchmark, or lack of statistical test.
   - Application gap: method exists but has not been validated in the user's domain.
   - Theory gap: mechanism, proof, convergence, stability, uncertainty, or causal explanation missing.

5. Position novelty.
   - Compare user contribution with the closest 5-10 papers.
   - State novelty in specific technical terms, not broad claims.
   - Classify contribution as algorithmic, experimental, theoretical, system, dataset, application, or integrative.

6. Write related work.
   - Start with the problem landscape.
   - Use thematic paragraphs.
   - End each theme by explaining what remains unresolved.
   - Close with a short paragraph that bridges to the user's method.

## Output Format

Provide:

- Search strategy and keyword groups.
- Literature matrix.
- 3-5 defensible research gaps with evidence.
- Closest-work comparison.
- Suggested contribution statement.
- Related-work draft or outline.

## Quality Checklist

- Covers direct competitors and foundational work.
- Distinguishes background, related work, and closest work.
- Does not claim "no one has studied this" unless the evidence is strong.
- Every gap is tied to papers, metrics, datasets, or technical constraints.
- The final novelty statement is narrow enough to survive peer review.