Replies: 1 comment 2 replies
-
see also the pre-print "The SAFE Procedure: A Practical Stopping Heuristic for Active Learning-Based Screening in Systematic Reviews and Meta-Analyses": https://psyarxiv.com/c93gq |
Beta Was this translation helpful? Give feedback.
2 replies
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
-
Afternoon everyone, I am using ASReview for a SR&MA and have 11,289 to screen after removing duplicates. Based on some of the supporting literature, 95% of these will be irrelevant. So, my question is, would you stop after screening the 5% (565 papers)? If so, what is the supporting citation to justify this for reviewers? This is with the understanding that ASReview will have moved the most relevant to the front of the screening process. My model is below, reference to their justification would also be appreciated.
Feature extraction: TF-IDF
Classifier: Naive Bayes
Query strategy: Maximum
Balance strategy: Dynamic resampling (double)
Thanks in advance
Ash
Beta Was this translation helpful? Give feedback.
All reactions