Implementasi Sistem Ujian Berbasis Online Pada Ujian Esai Bahasa Indonesia

Abstract

Classification of news manually impossible. Automatic classification lot to do with the naïve Bayes algorithm, but the number of words that many features can reduce the accuracy of the classification. This study aims to determine the application, influence and value of recall, precision, f-measure and accuracy of election Chi Square feature of the performance Naïve Bayes algorithm to automatically classify news text. In this study feature selection techniques applied by Chi Square in Naïve Bayes algorithm. Data were taken from as many as 1350 pieces www.kompas.com as training data and 150 as test data. Testing is done by classifying feature election news without Chi Square and classifying news by applying the Chi Square feature selection with significance level α 0:05, 0:01, 0.005, and 0.001. Next will be evaluated by the evaluation method of recall, precision, f- measure and accuracy. Automatic classification of news without selecting a feature that is done, the result recall 96.67%, 96.75% precision, f-measure 96.68% and 96.67% accuracy. While the classification of news with feature selection using the chi square on the real level α 0:05, 0:01, 0005, and 0001 obtained the same result, namely 98% recall, 98% precision, f-measure 97.99%, and accuracy 98%. From these results, it is known that the selection of the feature using the chi square can affect the performance Naïve Bayes algorithm to automatically classify news.