OPINION MINING PADA REVIEW BUKU MENGGUNAKAN ALGORITMA NAÏVE BAYES

Abstract

Abstract — In the era of widespread use of the internet today, the
number of consumers who wrote the opinion and experience of
online continues to increase. Read the review as a whole can be
time consuming, however, if only a few reviews that read, then the
evaluation will be biased. Sentiment analysis aims to address this
problem by automatically classifying user review be positive or
negative opinion. Naïve Bayes classifier is a popular machine
learning techniques for text classification, because it is very simple,
efficient and has a good performance in many domains. However,
Naïve Bayes has the disadvantage that is very sensitive to feature
too much, resulting in a classification accuracy becomes low.
Therefore, in this study used the integration method of feature
selection, namely Information gain and Genetic algorithm in order
to improve the accuracy of Naïve Bayes classifier. This research
resulted in the classification of the text in the form of positive or
negative review of the book. Measurement is based on the accuracy
of Naive Bayes before and after the addition of feature selection
methods. The evaluation was done using a 10 fold cross validation.
While the measurement accuracy is measured by confusion matrix
and ROC curves. The results showed an increase in the accuracy of
Naïve Bayes from 78.50% to 84.50%.