Performance of classifiers on newsgroups using specific subset of terms

Abstract

Text Classification plays a vital role in the world of data mining and same is true for the classification algorithms in text categorization. There are many techniques for text classification but this paper mainly focuses on these approaches Support vector machine (SVM), Naïve Bayes (NB), k-nearest neighbor (k-NN). This paper reveals results of the classifiers on mini-newsgroups data which consists of the classifies on mini-newsgroups data which consists a lot of documents and step by step tasks like a listing of files, preprocessing, the creation of terms(a specific subset of terms), using classifiers on specific subset of datasets. Finally, after the results and experiments over the dataset, it is concluded that SVM achieves good classification output corresponding to accuracy, precision, F-measure and recall but execution time is good for the k-NN approach.