Table 1.
Summary of state-of-the-art models.
Fig 1.
Proposed Framework of this research includes data collection, pre-processing, model training, and testing.
Fig 2.
Dataset categories distribution.
Fig 3.
Distribution of (a) word counts and (b) character length.
Fig 4.
Word cloud before pre-processing.
Fig 5.
Word cloud after pre-processing.
Fig 6.
Global feature extraction process.
Fig 7.
TF-IDF (term frequency-inverse document frequency).
Table 2.
Hyperparameters for different models and components.
Table 3.
System specifications and configuration.
Fig 8.
Confusion matrix of (a) CNN and (b) LSTM.
Fig 9.
Confusion matrix of (a) Decision Tree and (b) Logistic Regression.
Fig 10.
Confusion matrix of (a) Naive Bayes and (b) Random Forest.
Fig 11.
Confusion matrix of (a) Ensemble and (b) Stacking Classifier.
Table 4.
Performance analysis of different combinations of models in our dataset.
Fig 12.
ROC curve of (a) CNN and (b) LSTM.
Fig 13.
ROC curve of (a) Decision Tress and (b) Logistic Regression.
Fig 14.
ROC curve of (a) Naive Bayes and (b) Random Forest.
Fig 15.
ROC curve of (a) Ensemble and (b) Stacking Classifier.
Fig 16.
Some highlighted words form word-cloud.