Klasifikasi Dokumen Menggunakan Metode Rocchio Dengan Pemilihan Fitur Menggunakan Chi-square

Abstract

The drawback of document classification technique of a classification system is the significant increase of efforts when the contents of each document are large. One way to improve the performance of the classification system is by implementing the feature selection technique. This research implemented the Rocchio method as the document classifier and Chi-square as the feature selection technique at the significant level of 0.05 and 0.001. The document collections utilized in this research was taken from Reuters-21578 with 20 categories. The result showed that the effectiveness of classification in terms of F-measure and accuracy was insignificant. The highest F-measure value of 72.57% was achieved at a significant level of 0.001, or increasing 2.28% if compared with the previous F-measure value before feature selection method was applied. At a significant level of 0.05, the F-measure value was 72.42%. The obtained classification accuracy with feature selection technique was 83.67% at a significant level of 0.001 and 83.96% at a significant level of 0.05. These classification accuracies were higher than that without feature selection technique of 83.08%.