| Sumario: | The Quranic oath is God's emphasizing the importance or truthfulness of a concept. Oaths are multifaceted, rich expressions, in which a single oath contains a line of meaning and a variety of aspects. This study proposes a new stylometric model for detecting apparent and narrative oaths. Toward this end, two types of application-specific features from a stylometric perspective--structural and content- specific features--were examined. The stylometric features were extracted, and a Bayesian network was constructed to model such features. The stylometric model of oaths was then evaluated through a series of machine-learning experiments using various classifiers: the Bayesian network, a decision tree, instancebased learning, and a neural network. These classification experiments focused on applying stylometric features in apparent and narrative oaths. The experiments covered two datasets: the entire Quran and the smaller dataset of Juz' 'Amma. The results led to two main conclusions. First, stylometric application-specific features are best used in their entirety--both structural-based and content-specific--rather than as two separate entities. Second, applying stylometric features was more significant in Juz' 'Amma, in which 40% of its surahs (chapters) contain oath statements. Finally, the stylometric model was extended for oath styles detection using three additional stylometric features--syntactic, character, and lexical, and it was analyzed using statistical approach.
|
|---|