Sentiment analysis on Indonesian stock investment application (IPOT) reviews using naive bayes algorithm and genetic algorithm as feature selection method

Stock investment application is getting more attention from the public. Nowadays the Internet has an important role in promulgating information to its users. There are a huge amount of data generated by internet users that has their sentiments and opinions. Sentiment analysis is an approach to identify and categorize the polarity of internet users’ opinions. This approach is advantageous for companies in order to know consumers opinions about their product, and it can be useful for the consumers to determine the best products based on people’s opinions. Naïve Bayes Algorithm is one of the methods that is usually used in sentiment analysis research, because its simplicity and efficiency. But, Naïve Bayes is also very sensitive in feature selection. In order to resolve the deficiency of Naïve Bayes Algorithm, many researches combined this method with different kind of feature selection methods. The objective of this research is to combine Naïve Bayes Algorithm with Genetic Algorithms as a feature selection. After the combination, the classification accuracy of Naïve Bayes increases from 91.00% into 94.25%.