Online News Classification Using Hybrid Approach Including ANN

With the vast growth for online information, the concept of text classification came out to be one of the research areas that can handle the data adequately. To locate the information from the huge database has become significant these days. This research work has dealt with the evaluation of the algorithm based on text classification techniques in the text for enhancing the classification rate quantity of the provided text. The study finds the drawbacks of the classification of the Indian portal new sites being retrieved from different media sources having varied categories, such as, sports, financial, entertainment, regional and politics. The unstructured documents are converted into structured information after the classification process. For the training of the system, there are different types of classifier are used like SVM (Support Vector machine) and NN (Neural network) for the comparative analysis. These methods can be known as the methods of feature weigh adjustment. After the training and extraction of the features, classification of the system is done based on some testing data. The classification of the proposed work is dependent on the training structure of classifiers. Parameters, namely error rate and accuracy have been measured to check the efficiency of the proposed system. It is being seen from the results that the accuracy for NN is 99.91%, for SVM is 99.83 and for base paper (naïve baye’s) is 92.33%. From obtained results the value of neural network comes out to be better than SVM.