Considerations for conducting systematic reviews: A follow‐up study to evaluate the performance of various automated methods for reference de‐duplication
Why this work is in the frame
A frame that forgets how it found something cannot be audited. These are the routes that admitted this work.
Bibliographic record
Abstract
Searching multiple resources to locate eligible studies for research syntheses can result in hundreds to thousands of duplicate references that should be removed before the screening process for efficiency. Research investigating the performance of automated methods for deduplicating references via reference managers and systematic review software programs can become quickly outdated as new versions and programs become available. This follow-up study examined the performance of default de-duplication algorithms in EndNote 20, EndNote online classic, ProQuest RefWorks, Deduklick, and Systematic Review Accelerator's new Deduplicator tool. On most accounts, systematic review software programs outperformed reference managers when deduplicating references. While cost and the need for institutional access may restrict researchers from being able to utilize some automated methods for deduplicating references, Systematic Review Accelerator's Deduplicator tool is free to use and demonstrated the highest accuracy and sensitivity, while also offering user-mediation of detected duplicates to improve specificity. Researchers conducting syntheses should take automated de-duplication performance, and methods for improving and optimizing their use, into consideration to help prevent the unintentional removal of eligible studies and potential introduction of bias to syntheses. Researchers should also be transparent about their de-duplication process to help readers critically appraise their synthesis methods, and to comply with the PRISMA-S extension for reporting literature searches in systematic reviews.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.806 | 0.772 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.005 | 0.001 |
| Bibliometrics | 0.001 | 0.003 |
| Science and technology studies | 0.001 | 0.000 |
| Scholarly communication | 0.002 | 0.000 |
| Open science | 0.002 | 0.000 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.001 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from it