| name | stem-literature-review |
| description | 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. |
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
-
Clarify the research object.
- Extract core problem, application domain, technical method, and evaluation metric.
- Separate "what problem is solved" from "what method is used".
-
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.
-
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.
-
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.
-
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.
-
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.
