News Sentiment Analysis Using Naive Bayes And Adaboost

Abstract

Most of the data that exists today is in the form of digital data. If we sell shares or write a book even sell products online, it always involves an electronic devices. Since the paper transactions is occurred in digital form, a lot of data are available to be analyzed. One of data or information in digital form is information about the news. Information provided by the news provider's website contains a variety of things such as economics, politics, sports, and the others. The news has a variety of interesting patterns to be analyzed. The pattern can be used to predict the sentiment which contained in a few words, phrases, or sentences from a paragraph of news. This research discusses about sentiment analysis to the word or phrase which is used as test data to produce several classes such as positive, negative, and neutral sentiment. The method that used to weighting is using TFIDF word, then labeling of sentences is using sign numbers multiplication. Next, training and testing are using Naive Bayes method with a combination of AdaBoost.