Review Paper Writing Skill#

Overview#

This skill helps you write a thorough, well-structured literature review — from topic definition and systematic search to formatted citations and a polished markdown draft. It covers:

• Topic framing with PICO framework
• Search strategy (Boolean, MeSH, wildcards, proximity)
• Multi-source academic search (arXiv, Semantic Scholar, OpenAlex, CrossRef, PubMed)
• DOI / arXiv ID validation and conflict detection
• Citation formatting (APA, MLA, IEEE, BibTeX, RIS, Markdown)
• Auto-generated literature review in Markdown

Quick Start#

1. Run a Search#

bash

python3 scripts/lit_search.py "large language model education" -n 20 --sources arxiv openalex

2. Validate DOIs#

bash

python3 scripts/lit_search.py "machine learning" --validate

3. Format Citations#

bash

python3 scripts/lit_search.py "BERT" -n 10 -f apa      # APA 7th
python3 scripts/lit_search.py "transformer" -n 10 -f bibtex  # BibTeX
python3 scripts lit_search.py "attention" -n 10 -f ris     # RIS

4. Filter Results#

FlagEffect--published-onlyExclude arXiv preprints--preprint-onlyOnly arXiv preprints--min-year 2020Earliest publication year--min-citations 50Minimum citation count--conflictsDetect cross-source metadata conflicts

5. Batch Mode (from file)#

Create papers.txt with one DOI/PMID/arXiv ID per line:

text

10.48550/arXiv.2303.08774
10.48550/arXiv.2205.11916
PMID: 34567890

Then:

bash

python3 scripts/lit_search.py file://papers.txt -f bibtex

Search Strategy Framework#

PICO Framework#

Use PICO to structure your research question:

ElementMeaningExampleP opulationTarget grouphigh school studentsI nterventionTreatment or methodmotivational interviewingC omparatorControl groupno interventionO utcomeMeasurable resultacademic performance

Convert to search terms:

text

P: student*
I: "motivational interviewing"
O: "academic achievement"
→ "motivational interviewing" AND student* AND "academic achievement"

Boolean Operators#

python

# AND: all terms must appear
"cognitive behavioral therapy" AND adolescent*

# OR: any term may appear
depression OR anxiety OR "mental health"

# NOT: exclude terms
animal NOT pet

# Parentheses for grouping
("cognitive behavioral therapy" OR CBT) AND adolescent*

Wildcards & Truncation#

python

student*   → student, students, student's, students'
wom?n      → woman, women

Proximity Search#

python

behavior N/2 change    # "behavior" within 2 words of "change"
"stress" ADJ3 "coping" # same sentence

MeSH (PubMed)#

python

"Motor Activity"[Mesh]
"Depression"[Mesh] AND "Therapy"[Mesh:Subheading]

Database-Specific Syntax#

python

# PubMed
"motivational interviewing"[Title/Abstract]
("Anxiety"[Mesh] OR "Anxiety Disorders"[Mesh]) AND 2020:2024[dp]

# PsycINFO (OVID)
exp Motivational Interviewing/
motivation interview*.tw

# Web of Science
TS=("motivational interviewing" AND school*)
PY=2020-2024

Source Comparison#

SourceTypeAPI KeyBest ForarXivPreprint❌ NoCS, ML, physics, statsSemantic ScholarPublished + preprintOptionalHigh-citation papersOpenAlexPublished❌ NoBroad cross-disciplinaryCrossRefPublished❌ NoAuthoritative DOI metadataPubMedPublished (biomedical)❌ NoMedicine, health, biology

Recommended combinations:

• arxiv + semantic_scholar → preprints + high-impact papers
• openalex + pubmed → comprehensive academic coverage
• arxiv + openalex + pubmed → full-spectrum search

Quality Tags#

SourcePeer ReviewedTagarXiv❌ No[arXiv]Journal / Conference✅ Yes[Published]PubMed✅ Yes[PubMed]

⚠️ arXiv papers are preprints — not peer-reviewed. Treat conclusions as provisional.

Citation Formats#

FormatStyleapaAPA 7th EditionmlaMLA 9th EditionieeeIEEEbibtexBibTeX (for LaTeX)risRIS (for EndNote/Zotero)markdownMarkdown with source tags

Search Workflow#

text

1. Define topic → PICO framework
   ↓
2. Build search blocks → synonyms, MeSH terms, wildcards
   ↓
3. Run multi-source search (parallel API calls)
   ↓
4. Deduplicate by DOI / arXiv ID
   ↓
5. Validate IDs (HEAD request → doi.org)
   ↓
6. Detect cross-source conflicts (title, year, authors)
   ↓
7. Filter → year, citations, source type
   ↓
8. Format → APA / MLA / BibTeX / ...
   ↓
9. Generate literature review (timeline, keywords, summary)

Configuration#

Edit scripts/lit_search.py to add your Semantic Scholar API key:

python

API_KEYS = {
    "semantic_scholar": "your-api-key"  # Get: https://www.semanticscholar.org/api-keys
}

Without a key, the free tier is used (1 req/sec). A key enables 1000 req/5 sec and avoids rate-limiting.

Output Examples#

Markdown output ( -f markdown):

text

[1] **Attention Is All You Need**
* Vaswani, A., Shazeer, N., Parmar, N., et al. (2017). *
arXiv: [1706.03762](https://arxiv.org/abs/1706.03762) [arXiv]
*Citations: 80000*

[2] **BERT: Pre-training of Deep Bidirectional Transformers**
* Devlin, J., Chang, M. W., Lee, K., & Toutanova, K. (2019). NAACL-HLT 2019*
DOI: [10.48550/arXiv.1810.04805](https://doi.org/10.48550/arXiv.1810.04805) [Published]
*Citations: 70000*

Generated review section:

text

# Literature Review

## Statistics
- **Total Papers**: 20
- [arXiv]: 8
- [Published]: 10
- [PubMed]: 2

## Timeline
- 2024: 5 papers
- 2023: 8 papers
- 2022: 4 papers
- 2021: 3 papers

Troubleshooting#

ProblemSolution"Rate limit" from Semantic ScholarAdd API key, or switch to OpenAlexNo resultsTry fewer / broader search termsToo many resultsNarrow date range, add AND filtersInvalid DOICheck format; try CrossRef validationarXiv ID not foundUse format YYMM.NNNNN (e.g., 2303.08774)