<![CDATA[The influence of this growing culture and lifestyle makes people pay more attention to their appearance. One of the factors that affect appearance is the condition of a person's facial skin. Each product used by consumers has different reactions from one consumer to another, thus making many consumers review the products they use. Reviews given by consumers can be used to measure the quality of a beauty product. However, the large number of reviews given makes the review grouping unable to be done manually and sentiment analysis must be done to group the reviews into several categories. One of the algorithms for classifying sentiment analysis is using the Naive Bayes Classifier method which is a simple method that has faster performance in training data, is easy to implement, and has high effectiveness. In the classification process, feature selection will be used using the N-gram algorithm and DF-Thresholding to reduce the dimensions of the features in the data. The purpose of this study is to determine the effect of DF-Thresholding algorithm on the accuracy of the Naive Bayes Classifier algorithm using the N-gram. The result showed a reduction of 16.312 features to 43 features and the highest accuracy value for bigram and unigram combination, which is 49%, precision is 0,23, recall is 0,26 and f-measure is 0,24.]]>
Copyrights © 2021
