Table 2 Summary of existing research studies employing ML models for clickbait detection.
From: Clickbait detection in news headlines using RoBERTa-Large language model and deep embeddings
Ref | Models | Dataset | Features | Results | Limitation |
|---|---|---|---|---|---|
RF | Clickbait Challenge 2017 | Text features | 82 | Relies only on top sixty features; lacks contextual semantic representation. | |
AdaBoost | Clickbait news | Text features | 91 | Focused on feature engineering; limited generalization beyond dataset. | |
SVM | online news outlets | Text features | 97 | Compared multiple algorithms but lacked interpretability and semantic analysis. | |
LR | BollyBAIT | Word embeddings | 95 | Achieved accuracy with minimal features; limited exploration of deep models. | |
NB | Clickbait headlines | non-word features | 89 | Focused only on Arabic dataset; domain-specific and lacks cross-lingual evaluation. | |
RF | Arabic news | FV-ANOVA | 92 | Lower accuracy with traditional ML models; limited contextual representation. | |
DT | Indonesia news | Text features | 71 | Expanded dataset but relied on conventional ANN/LSTM; limited interpretability. | |
RF | Turkish clickbait | Word2vec | 97 | Compared to NB and LSTM, it lacks robust dataset diversity. | |
NB | Clickbait news | Text feature | 96 | Limited exploration of deep models. |
