Non-Probabilistic K-Nearest Neighbor for Automatic News Classification Model with K-Means Clustering

The news classification is the branch of text classification or text mining. The researchers have already done a lot of work on the text classification models with different approaches. The news works has to be classified in the form of various categories such as sports, political, technology, business, science, health, regional and many other similar categories. The researchers have already worked with many supervised and unsupervised methods for the purpose of news classification. The supervised models have been found more efficient for the purpose of news classification. The k-means algorithm has been used for the classification of the keywords into the multiple groups. The k-nearest neighbor (kNN) classification algorithm has been utilized to estimate the category of the news in the processing. The proposed model has been recorded with the average accuracy of the 93.28% obtained after averaging the accuracy of all test cases, which higher than the previous best performer naïve bayes and SVM based news classifier, which has posted nearly 83.5% of accuracy for classifying the news data. The proposed model has been tested with the 91%, 95%, 90% and 97% of the accuracy over the input test cases of S1, S2, S3 and S4 respectively, which higher than all of the existing models. Hence the proposed model can be declared as the better solution than the previous classification models. KEYWORDS—News classification, k-nearest neighbor, k-means classification, support vector machine, N-gram analysis.