Combining Probabilistic Classifiers for Text Classification

Abstract

In the realm of machine learning, probabilistic models are widely regarded as some of the best available. Very little research has been
done to evaluate the performance of two or more classifiers used in conjunction in the same classification task, despite the fact that famous
probability classifiers show very excellent performance when used separately in a particular classification task. In this study, we employ
two probability strategies for document classification: the naïve Bayes classifier and the Maximum Entropy model. Then, we merge the
two sets of findings using two different operators—Max and Harmonic Mean—to boost the categorization performance. Results from an
evaluation conducted on the "ModApte" subset of the Reuters-21578 dataset demonstrate that the suggested technique improves final
evaluation accuracy significantly.