Complement Naive Bayes Classifier for Sentiment Analysis of Internet Movie Database

Sentiment analysis (SA), often known as opinion mining, is the subjective examination of a written text. Moreover, SA is a critical technique in today's artificial intelligence (AI) field for extracting emotional information from huge amounts of data. The study is based on the Internet Movie Database (IMDB) dataset, which comprises movie reviews and the positive or negative labels that are connected with them. Our research experiment's objective is to identify the model with the best accuracy and the most generality. Text preprocessing is the first and most critical phase in a Natural Language Processing (NLP) system since it significantly impacts the overall accuracy of the classification algorithms. The experiment implements the Term Frequency-Inverse Document Frequency model (TFIDF) to feature selection and extractions. The following classifiers are used in this work: Linear Model and Naive Bayes. Besides, we explore the possible options of loss functions such as square_hinge, huber, modified_huber, log, epsilon_insensitive, perceptron, and modified_huber. ComplementNB achieves the highest accuracy, 75.13%, for both classification reports based on our experiment result.