Author(s): Çiğdem AYTEKİN, Cem Sefa SÜTCÜ, Umut ÖZFİDAN
Most of the data available today are text based. This necessitates developing some methods for their analysis. Because inspecting these text is very difficult, even impossible most of the time. Necessity of extracting knowledge from text data has triggered works about automatically extracting knowledge out of text data and text classifications methods have emerged. But since text data are not structural, their analysis are different than traditional machine learning applications. In this study, by selecting a sample from customer comments in a firm’s database, a decision tree model is constructed which can assign these comments into complaint-request-acknowledgement classes. Algorithm is based on entropy and knowledge gain calculations. This way, first attributes -words- that can represent customer comments have extracted and by defining nodes class labels.
The Journal of International Social Research received 8982 citations as per Google Scholar report
