Determination of e-commerce sites by text mining

In this thesis, an application which decide whether or not the pages specified by the user are electronic commerce (e-commerce) sites by analyzing the contents of web sites and by using text mining techniques is developed. The application is designed to be used in special studies and aims to facilitate users to find ecommerce sites. In this study, the data was collected from different sources to be used in text classification and prepared for useage. A static and a dynamic study were performed. Different training sets are prepared based on these studies and k-NN Nearest Neighbour and Naïve Bayes classification algorithms are used on these sets. And then obtained results are compared. Algorithm that gives better results is selected and a desktop application with the Java programming language is developed for usage.