Analisis Kinerja Smoothing pada Naive Bayes untuk Pengkategorian Soal Ujian

Abstract

Cognitive domain taxonomy bloom is a reference to classify examination based on difficulty levels on education world, the result of those categories will be used to compile a variety of exam questions. In this research, Naive bayes classifier will be used to categorize the text about biology exam for high school. Chi-square is feature selection method that will be used to remove an unuse features on examination, and increase speed of process text categorization. In addition, naive bayes classifier is a method that causing missclassification result of categorization text, this case will be happen if testing data is not found in training data, so we need another method to minimize it, that method called by smoothing method. In this research we will test perfomance and impact of smothing method for naive bayes classifier and chi-square as feature selection method. The smoothing methods to be compared on this research are : Laplace, Dirichlet and Two Stage smoothing. DOI: https://doi.org/10.26905/jtmi.v4i2.2080

Keywords

• bloom’s taxonomy
• smoothing
• naive bayes
• chi-square, feature selection
• examination