Table 6 Different researches classification method and results.
From: An open automation system for predatory journal detection
Research | Dataset | F1-score |
|---|---|---|
Al-Matham & Al-Khalifa (2017) | IsPredatory publishers' database which contains 1000 + entries of predatory publishers |  |
Bedmuth et al. (2020) | 6,268 articles from OMICS (as predatory) and 34,763 articles from BMC (as non-predatory) | Engineering Area Naïve Bayes:0.89 Random Forest: 0.83 Decision Tree: 0.71 Biomedical Area Naïve Bayes:0.96 Random Forest: 0.93 Decision: 0.90 |
Adnan et al. (2018) | 200 websites for training 200 websites for testing (Beall’s list as predatory, directory of open access journals lists as non-predatory) | Heuristic features Naïve Bayes:0.95 SVM: 0.98 KNN(k = 3): 0.94 NWF Naïve Bayes: 0.89 SVM:0.96 KNN(k = 3): 0.93 |
Our results | 1,636 websites for training 410 websites for testing (Predatory Journal Lists were collected from updated Beall’s and the Stop Predatory Journals lists. Legitimate journal list data were collected from the Berlin Institute of Health (BIH) Quest website) | NWF Gaussian Naïve Bayes: 0.89 SVM:0.952 Random Forest: 0.98 SGD:0.972 KNN(k = 4): 0.945 |
