An efficient machine Learning Bayes Sentiment Classification method based on review comments

Abstract

A major concern while incorporating semantic knowledge bases for opinion mining is that the words selected does not solve attribute relevancy and could not ground positive and negative usage of ambiguous terms. These concerns often make it difficult to classify the opinion words from user review comments. This paper presents a novel method called Machine Learning Bayes Sentiment Classification (MLBSC) to improve the classification accuracy by forming classes (i.e., positive, neutral and negative) based on the extracted words from user review comments. Initially, related opinion words are organized for its semantic equivalence of sentiments based on prior training list (i.e. using extracted words). Then probabilistic Bayes classifiers are applied on semantic opinion words to evaluate sentiment class label. The sentiment class labels are trained for positive, neutral and negative sentiments with the user review comments. The method MLBSC is evaluated for customer review data sets from research repositories. The MLBSC method produces attribute relevancy and economically significant gains for customers and performs better out of sample based on review comments. An intensive and comparative study shows the efficiency of these enhancements and shows better performance in terms of classification accuracy, size of classes, density of class label, execution time for class generation.