A Comprehensive Analysis of Facebook Bangla Comments to Examine Public Views of Male & Female Bangladeshi Cricket Players Using Machine Learning & Deep Learning Algorithms

Social media allows anyone to express their feelings, but sometimes people fail to follow community standards and cross the boundaries of their limits. In our paper, we aim to understand the overall scenario of people’s opinions about twelve male and female Bangladeshi cricketers, and we have developed a Bangla text-based dataset that contains 5702 Bangla comments from Bangladeshi male cricketers and 1381 Bangla comments from Bangladeshi female cricketers verified Facebook profiles. To analyze these comments, we categorized them into three polarities such as positive, negative, and neutral. In addition, we collected seven reactions to these comments. After preparing our data, we applied nine machine and deep learning algorithms consisting of K-Nearest Neighbors (KNN), Adaptive Boosting (AdaBoost), Linear Support Vector Machine (SVM), Artificial Neural Network (ANN), Extreme Gradient Boosting (XGBoost), Logistic Regression (LR), Random Forest (RF), and Decision Tree (DT). Our multiclass classification algorithm, consisting of TF-IDF vectorizer alongside uni-gram, bi-gram, and tri-gram followed by KNN, LR, and RF demonstrates the topnotch performance in 90-10 and 80-20 split respectively while RF performed better in 70-30 split. We also employed word2vec for nine algorithms including Convolutional Neural Network (CNN) which obtained the best performance in 70-30 split.