Comparative study on using artificial neural networks for text categorization

This research provides a comparative study on using Artificial Neural Networks for text categorization. In a text categorization model using an artificial neural network as a classifier, performance can be an issue if the neural network is trained using the raw feature space since textural data has a very high dimension feature space. We have used five dimensionality reduction techniques to reduce the feature space into an input space of a much lower dimension for the neural network classifier and then observed its effectiveness as a text classifier. To test the effectiveness of the proposed approach, experiments have been conducted using the “20-Newsgroups” data set. The learning machines that have been used in the work are “Radial Basis Functions (RBF)” and “Support Vector Machines (SVM)”. The performance of a Naïve Bavesian Classifier has been used as a benchmark for the comparative study.; This work also touches the Naïve Bayesian Classifier's performance with respect to its saturation point and unique feature set.