ANALISIS SENTIMEN MENGGUNAKAN METODE NAÏVE BAYES CLASSIFIER DENGAN SELEKSI FITUR CHI SQUARE

Abstract

Sentiment analysis is the computational study of opinions, sentiments, and emotions expressed in texts. The basic task of sentiment analysis is to classify the polarity of the existing texts in documents, sentences, or opinions. Polarity has meaning if there is text in the document, sentence, or the opinion has a positive or negative aspect. In this study, classification of the polarity in sentiment analysis using machine learning techniques, that is Naïve Bayes classifier. Criteria for text classification decisions, learned automatically from learning the data. The need for manual classification is still required because training the data derived from manually labeling, the label (feature) refers to the process of adding a description of each data according to its category. In the process of labeling, feature selection is used and performed by chi-square feature selection, to reduce the disturbance (noise) in the classification. The results showed that the frequency of occurrences of the expected features in the true category and in the false category have an important role in the chi-square feature selection. Then classification breaking news by Naïve Bayes classifier obtained an accuracy of 83% and a harmonic average of 90.713%.