Abstract:
In the early days, very few studies were effectuated in Bengali Languages as well as its functional
sentence. However, studies on Bengali have grown exponentially due to structural diversity.
Inspired by these studies, the classification of the Bengali functional sentences was completed with
machine learning methods for classifying sentences. The study looked at three different forms of
Bengali functional sentences: assertive, interrogative, and exclamatory. Thus, The study's major
goal is to categorize the sentence and compare the rate of accuracy to determine the optimal model.
The dataset has been properly collected, classified, and processed to avoid conflicts. Some
conventional machine learning (ML) classifiers such as Decision Tree (DT), Naive Bayes (NB),
Support Vector Machine (SVM), Random Forest (RF), K-Nearest Neighbor (KNN), and Extreme
Gradient Boosting(XGB) have been applied to compare classification accuracy rates. Parameters
such as precision, recall, F1-score, support, and confusion matrix were calculated for comparison.
The comparison proved that the performance of RF, SVC, and XGB classifiers is better than Naive
Bayes and Decision Tree classifiers. A notable enigma is that the RF algorithm implemented the
highest attainment value with 75.38% accuracy which is the ordinary performance of such datasets.
