Machine learning-based new approach to films review

The main purpose of Sentiment Analysis (SA) is to derive useful insights from large amounts of unstructured data compiled from various sources. This analysis helps to interpret and classify textual data using different techniques applied in machine learning (ML) models. In this paper, we compared simple and ensemble ML methods as classifiers for SA: Random Forest, K-Nearest Neighbor, Artificial Neural Network, Gradient Boosting, Support Vector Machine (SVM), AdaBoost, Extreme Gradient Boosting, Decision Tree, Light GBM, Stochastic Gradient Descent and Bagging. For this, we considered a test set database of 50,000 movie reviews, of which 25,000 were rated positive and 25,000 negatives. We have chosen 20,000 words that have an impact on the feelings of the documents. This work aims to propose a new rating prediction approach based on a textual customer review. We consider term frequency characteristics and term frequency-inverse document frequency from the large-scale and serial trials to compare the results obtained by various classifiers using feature extraction techniques. For the decision phase, we applied the Fuzzy Decision by Opinion Score Method, one of the most recent methods for multi-criteria decision-making. To evaluate and quantify the performance of the different ML methods we considered, we apply six standard measures namely precision, accuracy, recall, F-score, AUC, and Kappa-measure. The results we obtained, at the end of the experimental work that we conducted, indicated that the SVM classier is the best with 88,333% as a precision rate followed by the FDOSM method, with 0.800 for the same measurement.