In this article, we will unveil a novel probabilistic graphic model (PGM) based on the BabelNet knowledge resource for the classification of Arabic text. Unlike the standard Bag-of-Words system for text classification, this new approach focuses on a formal probabilistic model in which each entity depicted by a graph communicates the conditional relation structure between the different named entities. This method could be used in situations where data is massive (big data) and the traditional solution cannot be accurate. The experimental results demonstrate that the graphical representation founded on the Support Vector Machine (SVM) exceeds the Naive Bayes (NB) for all measurements.
Volume 12 | 07-Special Issue
Pages: 1241-1250
DOI: 10.5373/JARDCS/V12SP7/20202224