Employing the K-Means Algorithm to Detect Regular Rules in the Structures of Infinitives of Three-Letter Verbs

Abstract

The study aimed to investigate the extent of regularity in the structures of infinitives of three-letter root verbs within consistent rules, utilizing the K-Means algorithm, a machine learning technique designed to cluster inputs into homogeneous groups. To achieve the study’s objective, a sample of 2,086 three-letter root verbs was collected, and their features were extracted from the “Dictionary of Three-Letter Verbs in Arabic”. The researcher adopted a descriptive-Inductive approach by describing and encoding the input features prior to processing, followed by an analysis of the results yielded by the algorithm. The study concluded with several findings, most notably: the identification of thirteen (23) consistent rules within the examined sample; the infinitive pattern "Fa'l" accounted for the largest share, comprising eleven (19) of these rules; and the demonstration of the significant influence of the Morphological Features on rule formulation. Based on these findings, the study offered key recommendations, including the necessity of separating feature values containing multiple possibilities to enhance rule regularity, and leveraging computational power to manage the multitude of detailed rules that exceed human cognitive capacity.