Discovering The Applicability of Classification Algorithms With Arabic Poetry
Mohammad Gharbat, Heba Saadeh, Reem Q. Al Fayez · 2019 IEEE Jordan International Joint Conference on Electrical Engineering and Information Technology (JEEIT) · 2019
The classification algorithms were developed to label the objects with the same characteristics into similar groups. In Arabic poetry, poems belong to different eras based on the time they were written in. Changes in time resulted in changes in the style of writing Arabic poetry. Hence, this study is testing the use of different classification algorithms to discover their ability to classify poems in their correct era. This process can be automated by studying the linguistic changes that happened to the Arabic poetry in terms of the words used in the hemistichs structure (part of poem line) without considering poems rhyme. In this paper, we studied applying different classification algorithms to discover poems written in the Abbasid and Andalusian eras. Data have been collected and classifiers have been trained on an overall dataset that contains 30,058 words from 10,895 poetic hemistich collected from both eras. The experiments obtained 70.50% accuracy rate using the support vector machine classifier when tested with a sample of random poem lines. The study showed that it is possible to distinguish, with decent accuracy, poems from different eras based on identifying discriminant features that can be used in the classification.