Klasifikasi Dokumen Teks Menggunakan Metode Support Vector Machine dengan Pemilihan Fitur Chi-Square.

Abstract

Increasing number of documents makes people more difficult to obtain the information which they desired. This problem requires text processing techniques to organize the documents in accordance with the categories. One of which is text classification. Text classification can organize document in accordance with predefined categories automatically (supervised machine learning). One popular method of text classification is support vector machines (SVM) that tries to find the best hyperplane in the input space. This algorithm is the best classification algorithm compared with other vector space classification method, namely Rocchio, k-nearest neighbor (KNN) and decision tree. This research measures the suitability of SVM for text classification and to prove whether the SVM is able to classify the documents in a linear separable manner. The final result shows that linear kernel and polynomial kernel in the SVM test produce the same accuracy value of 96.3504% and testing the RBF kernel produces accuracy of 95.6204% for classification of text documents using chi-squared feature selection