Analisis Sentiment Tweets Berbahasa Sunda Menggunakan Naive Bayes Classifier dengan Seleksi Feature Chi Squared Statistic

Abstract

At present the development of the use of social media in Indonesia is very rapid, in Indonesia there are a variety of regional languages, one of which is the Sundanese language, where some people especially those living in West Java use Sundanese language to express comments, opinions, suggestions, criticisms and others in social media. This information can be used as valuable data for individuals or organizations in decision making. The huge amount of data makes it impossible for humans to read and analyze it manually. Sentiment analysis is the process of classifying opinions, analyzing, understanding, evaluating, emotions and attitudes towards a particular entity such as individuals, organizations, products or services, topics, events, in order to obtain information. The purpose of this research is the Naїve Bayes Classifier (NBC) classification algorithm and Feature Chi Squared Statistics selection method can be used in Sundanese-language tweets sentiment analysis on Twitter social media into positive, negative and neutral categories. Chi Square Statistic feature test results can reduce irrelevant features in the Naïve Bayes Classifier classification process on Sundanese-language tweets with an accuracy of 78.48%.