How to Use This Library
66,671 non-retracted papers, organized, scored, and ready for manuscript writing.
Quick Access Guide
| What you need to do | Where to go |
|---|---|
| Import the full reference library into EndNote | lit_review/master/master_library.ris -- 66,671 papers |
| Review only the highest-priority papers | lit_review/master/high_priority.csv -- Tier A + top Tier B |
| See which themes each paper belongs to | lit_review/master/theme_tags.csv |
| Start writing a manuscript section | lit_review/themes/theme_XX_*/section_notes.md |
| Browse papers for a specific theme | lit_review/themes/theme_XX_*/papers.md -- annotated markdown |
| Check for retracted papers before citing | lit_review/metadata/retracted_flagged.csv |
| Understand deduplication | lit_review/metadata/dedup_report.md |
| See exact search queries and hit counts | lit_review/themes/theme_XX_*/queries_run.md |
| Re-run or extend the pipeline | lit_review/scripts/ -- Python engine |
Understanding the Tier System
| Tier | Score Range | What it means | How to use it |
|---|---|---|---|
| A | >= 0.7 | Must-cite. Anchor papers, high-impact journals, heavily cited. | Cite these in the manuscript. They define the field. |
| B | 0.4 -- 0.7 | Strong supporting evidence. Good journals, relevant findings. | Use selectively to support specific claims. |
| C | < 0.4 | Background reference. May be tangentially relevant. | Consult if you need additional evidence for a specific point. |
Re-Running the Pipeline
cd lit_review/scripts
# Install dependencies
pip install requests lxml pandas tqdm rispy python-Levenshtein
# Run a single theme
python run_theme.py theme_01
# Generate synthesis notes for all themes
python synthesize.py all
# Build the master library
python build_master.py
All API responses are cached under lit_review/cache/. Re-runs use cached data automatically.