Fig 1.
Social media user distribution in Bangladesh (2023).
Table 1.
Related publications along with boundaries.
Fig 2.
Workflow of our proposed model.
Fig 3.
Percentage of comments for each class.
Fig 4.
Word cloud of dataset.
Table 2.
Numeric features of the dataset.
Fig 5.
Density plot of comments length before and after preprocessing the dataset.
Fig 6.
Original and processed text of Bengali text dataset.
Fig 7.
Example of CV & TF-IDF transformation representation of a comment.
Fig 8.
Example of tokenization and padding representation of a comment.
Fig 9.
Arrangement of the stacking ensemble.
Fig 10.
Arrangement of the voting ensemble.
Table 3.
ANOVA and chi2 tests results.
Table 4.
Comparative overview of all DL and ML algorithms parameters in used in this study.
Fig 11.
Results of training as well as validation: The matrix of confusion of all ML models.
(a) Stacking, (b) AdaBoost, (c) Bagging, (d) K-NN, (e) LR, (f) MLP, (g) RF, (h) SGD, (i) Voting, (j) XGBoost.
Fig 12.
Confusion matrix and ROC curve of hybrid stacking model with binary classifications.
(a) Confusion Matrix, (b) ROC curve.
Fig 13.
All ML models ROC curve over every class.
(a) Adaboost, (b) Bagging, (c) K-NN, (d) LR, (e) MLP, (f) RF, (g) SGD, (h) Voting, (i) XGBoost, (j) Stacking.
Table 5.
All ML models AUC score list.
Fig 14.
The collection of data assessment’s accuracy, precision, recall, and F1 score of all ML models.
Table 6.
All ML models classification reports of precision, recall, F1 score, and accuracy.
Fig 15.
Results of training as well as validation: The matrix of confusion of all ML models.
(a) BiGRU, (b) CLSTM, (c) CNN, (d) DNN, (e) RNN.
Fig 16.
All DL models ROC curve over every class.
(a) BiGRU, (b) CLSTM, (c) CNN, (d) RNN, (e) DNN.
Table 7.
AUC score comparison table for all the deep learning algorithms.
Fig 17.
The collection of data assessment’s accuracy, precision, recall, and F1 score.
Table 8.
All DL models classification report of precision, recall, F1 score and accuracy.
Fig 18.
Classification report of proposed hybrid model.
Table 9.
Comparison of the proposed model with previous investigations.
Fig 19.
Web view of our system.