alterlab-citation-mgmt

Citation Management

Overview

Manage citations systematically across the research and writing process: search academic databases (Google Scholar, PubMed), extract accurate metadata from multiple sources (CrossRef, PubMed, arXiv, DataCite), validate citation information, and generate properly formatted BibTeX. Critical for citation accuracy, avoiding reference errors, and reproducible research. Integrates with alterlab-literature-review.

When to Use This Skill

Use when:

Searching for specific papers on Google Scholar or PubMed
Converting DOIs, PMIDs, or arXiv IDs to properly formatted BibTeX
Extracting complete metadata (authors, title, journal, year, etc.)
Validating existing citations for accuracy or checking for duplicates
Cleaning, sorting, and formatting BibTeX files
Finding highly cited / seminal papers in a field
Building a bibliography for a manuscript or thesis

Does NOT Trigger

Scenario	Use Instead
Proving each reference actually exists, catching hallucinated/fabricated citations, or retraction checks	`alterlab-citation-verifier`
Reading or editing a Zotero library (items, collections, tags, attachments)	`alterlab-pyzotero`
A full systematic review — multi-database search, PRISMA screening, thematic synthesis	`alterlab-literature-review`
Mapping a citation/co-citation network around seed papers	`alterlab-citation-graph`
Writing manuscript prose around the references	`alterlab-scientific-writing`

Core Workflow

Citation management is a five-phase pipeline. Each phase maps to a script (full CLI in references/script_cli_reference.md) and a deep-dive reference.

Phase 1 — Paper discovery

Search Google Scholar (broad, cross-discipline) and PubMed (biomedical, MeSH-indexed).

python scripts/search_google_scholar.py "CRISPR gene editing" --limit 50 --output results.json
python scripts/search_pubmed.py "Alzheimer's disease treatment" --limit 100 --output alz.json

Operators, MeSH/field tags, complex queries, and high-impact-paper heuristics: references/search_strategy_reference.md, references/google_scholar_search.md, references/pubmed_search.md.

Phase 2 — Metadata extraction

Convert any identifier (DOI, PMID, arXiv ID, URL) to complete metadata. CrossRef is the primary DOI source (extract_metadata.py); PubMed E-utilities and the arXiv API cover PMIDs and arXiv IDs. doi_to_bibtex.py uses doi.org content negotiation, which also resolves DataCite DOIs (datasets, software).

python scripts/doi_to_bibtex.py 10.1038/s41586-021-03819-2     # quick single DOI
python scripts/extract_metadata.py --pmid 34265844             # any identifier type
python scripts/extract_metadata.py --input identifiers.txt --output citations.bib  # batch

Sources and extracted fields: references/metadata_extraction.md and the Metadata Sources section of references/search_strategy_reference.md.

Phase 3 — BibTeX formatting

Generate clean, standardized entries. Common types: @article, @book, @inproceedings, @incollection, @phdthesis, @misc. Protect title capitalization with {}, use -- for page ranges, include a DOI for modern publications.

python scripts/format_bibtex.py references.bib --deduplicate --sort year --descending \
  --output clean_references.bib

Entry types, required fields, and formatting rules: references/bibtex_formatting.md.

Phase 4 — Validation

Verify required fields are present, data is consistent, and there are no duplicates. DOI resolution is network-bound and OFF by default — add --check-dois to actually hit doi.org/CrossRef (slow). validate_citations.py reports only; it does not rewrite the file. To apply fixes (page dashes, author separators, dedup), run format_bibtex.py.

python scripts/validate_citations.py references.bib --check-dois \
  --report validation.json --verbose

Validation criteria and report format: references/citation_validation.md and the Validation Checks summary in references/search_strategy_reference.md.

Phase 5 — Integration with writing

Export validated BibTeX into a LaTeX manuscript (\bibliography{final_references}). Pairs with alterlab-literature-review (search/synthesis), alterlab-scientific-writing (manuscript references), and alterlab-venue-templates (style-specific formatting). Before submission, run alterlab-citation-verifier on the final .bib to confirm every entry exists — this skill checks fields and formatting, not existence.

End-to-end worked examples (build a bibliography, convert a DOI list, clean a messy .bib, find seminal papers): references/example_workflows.md.

Best Practices

Search: start broad then narrow; use multiple sources; leverage "Cited by" for seminal papers; document queries and dates.

Metadata: prefer DOIs (most reliable, best CrossRef metadata); verify author names, venue, and year; handle preprints (use the published version when one exists); keep author-name and journal-abbreviation formatting consistent.

BibTeX quality: meaningful citation keys (FirstAuthor2024keyword); protect title capitalization with {}; remove redundant fields; validate syntax regularly; organize per project and merge carefully to avoid duplicates.

Validation: validate early and often; fix broken DOIs and missing fields promptly; manually review critical citations.

Common Pitfalls

Single-source bias → search multiple databases.
Accepting metadata blindly → spot-check against originals.
Ignoring DOI errors → validate before submission.
Inconsistent formatting → standardize with format_bibtex.py.
Duplicate entries → use duplicate detection.
Missing required fields → validate for completeness.
Outdated preprints → update to the published version.
Special-character issues → escape or use Unicode in BibTeX.
No final validation → always run validation as the last check.
Manual BibTeX entry → always extract from metadata sources.

Index of Bundled Resources

References (`references/`)

script_cli_reference.md — full per-script features and CLI usage
example_workflows.md — four end-to-end worked workflows
search_strategy_reference.md — high-impact heuristics, Scholar operators, MeSH/field tags, metadata sources
google_scholar_search.md — complete Google Scholar search guide
pubmed_search.md — PubMed and E-utilities API documentation
metadata_extraction.md — metadata sources and field requirements
citation_validation.md — validation criteria and quality checks
bibtex_formatting.md — BibTeX entry types and formatting rules

Scripts (`scripts/`)

search_google_scholar.py, search_pubmed.py, extract_metadata.py, validate_citations.py, format_bibtex.py, doi_to_bibtex.py.

Assets (`assets/`)

bibtex_template.bib (example entries for all types), citation_checklist.md (QA checklist).

External Resources

Google Scholar: https://scholar.google.com/ · PubMed: https://pubmed.ncbi.nlm.nih.gov/
CrossRef API: https://api.crossref.org/ · PubMed E-utilities: https://www.ncbi.nlm.nih.gov/books/NBK25501/
arXiv API: https://arxiv.org/help/api/ · DataCite API: https://api.datacite.org/
MeSH Browser: https://meshb.nlm.nih.gov/search · DOI Resolver: https://doi.org/

Dependencies

# All network scripts (CrossRef, doi.org, E-utilities, arXiv)
uv pip install requests
# Optional: Google Scholar search (search_google_scholar.py). scholarly scrapes Scholar,
# which has no official API and rate-limits or CAPTCHAs automated traffic.
uv pip install scholarly

format_bibtex.py and validate_citations.py (without --check-dois) need only the standard library. Set NCBI_EMAIL (and optionally NCBI_API_KEY) for PubMed.

Summary

This skill provides search (Scholar + PubMed), automated metadata extraction (DOI/PMID/ arXiv/URL), citation validation (DOI verification + completeness), BibTeX formatting and cleaning, and quality assurance — all reproducible through documented search and extraction methods. Use it to maintain accurate, publication-ready bibliographies.