Multinomial Optimization of Naïve Bayes Through the Implementation of Particle Swarm Optimization

Abstract

Sentiment analysis is widely used in cases of text processing and comments. One of the case studies is about the analysis of a hotel review by the public. The method used in analyzing a sentiment from comments or reviews of a hotel is the Naïve Bayes Classifier. One that can be used is the Multinomial Naïve Bayes method. In improving the results of the accuracy of the method required an optimization method. There are many optimization methods that can be applied to algorithms in sentiment analysis case studies. One well-known method is Particle Swarm Optimization (PSO). This study aims to determine the effect of PSO optimization on the Multinomial Naïve Bayes algorithm in the case of sentiment analysis. From the results of optimization and model testing, the highest accuracy was obtained in the Multinomial Naïve Bayes test with PSO optimization as hyperparameter tunning and feature selection of 97%.