Authors : Wareesa Sharif, Noor Azah Samsudin, Mustafa Mat Deris and Shamsul Kamal Ahmad Khalid
Abstract: Since, electronic documents are dramatically increasing therefore document classification becomes a very important task to organise information automatically. Text documents are a high dimensional data that create difficulties in classification task. Consequently, various feature ranking techniques are used to reduce the dimensionality of the text data. Features can be selected by using document frequency and term frequency techniques. In term frequency, few researchers have worked on term frequency while keeping the property of document frequency. Little attention has been paid to compare these techniques. In this study, we describe issues of feature ranking techniques and classification document labelling problem. This study also present and discussed experimental result of feature ranking technique with presence or absence of term and term frequency (term count) in document classification problem. The result shows that redesigning of term into term count with document frequency could lead to better classification accuracy than that of term frequency and document frequency separately.
Wareesa Sharif, Noor Azah Samsudin, Mustafa Mat Deris and Shamsul Kamal Ahmad Khalid, 2018. A Technical Study on Feature Ranking Techniques and Classification Algorithms. Journal of Engineering and Applied Sciences, 13: 7074-7080.