Загрузка...

Citation Dedup in 60 seconds: 15-30 percent of records are duplicates. #Shorts

PubMed says yes. Embase says yes. Cochrane CENTRAL says yes. The same paper is in your library three times. Pool the four-database search uncritically and 20 percent of your "records" are duplicates. Smart deduplication removes them with audit trail intact.

How the matching works:
- Exact match on DOI (highest confidence)
- Exact match on PMID
- Fuzzy match on title (Levenshtein distance threshold, configurable)
- Author + year + journal triangulation for near-matches
- Bramer 2016 Journal of the Medical Library Association validated algorithm

When records merge, the tool preserves the union of metadata:
- Richest abstract (longest non-empty)
- All present DOIs, PMIDs, and accession numbers
- Per-database source provenance (so you know it came from PubMed AND Embase)
- All keyword and indexing terms

Manual review queue:
- Borderline matches (similarity score between thresholds) shown side-by-side
- Reviewer accepts or rejects each merge
- Audit log preserved for PRISMA-S reporting

Workflow position: use AFTER searching all databases, BEFORE screening. A typical four-database search has 15-30 percent duplicates - removing them upfront saves your reviewers hours.

Export: deduplicated RIS or CSV with source-database column, ready for the screening module.

Why this matters: Cochrane Handbook chapter 4.5 mandates explicit reporting of dedup methodology in the PRISMA flow diagram. Most reviewers do it informally; this tool makes it rigorous.

Used in Synthesis course Module 3: deduplication workflow.

Runs offline. Your library stays on device. MIT licensed, open source.

One of 71 tools in the allmeta evidence-synthesis suite.

Live: https://mahmood726-cyber.github.io/allmeta/citation-dedup/index.html
Catalogue: https://mahmood726-cyber.github.io/allmeta/
Synthesis Courses (free, 26 modules × 12 languages): https://mahmood726-cyber.github.io/synthesis-courses/
Evidence Reversal series: at meta-analysishtml

#MetaAnalysis #Deduplication #PRISMA #SystematicReview #Cochrane #Bramer #EBM #Shorts #YTShorts

Видео Citation Dedup in 60 seconds: 15-30 percent of records are duplicates. #Shorts канала 786-MIII Meta-analysis
Яндекс.Метрика
Все заметки Новая заметка Страницу в заметки
Страницу в закладки Мои закладки
На информационно-развлекательном портале SALDA.WS применяются cookie-файлы. Нажимая кнопку Принять, вы подтверждаете свое согласие на их использование.
О CookiesНапомнить позжеПринять