Exploring Authorship Differentiation in the Quran and Hadith Using Deep Learning and Machine Learning Approaches

A mysterious enigma about who wrote the Quran led to several authorship attribution analyses on the ancient book. Given the outperformance of deep learning over conventional ML techniques, a new idea arises suggesting the use of this recent approach to try solving this enigma.That is, in this research work, we conduct an authorship differentiation analysis between two Arabic religious texts: the holy Quran and Hadith. The primary objective is to ascertain whether the Quran and Hadith could have been authored by the same Author. The comprehensive textual corpus, comprising the Holy Quran and a verified portion of the Hadith, is divided into text segments of 500 words each.We propose and evaluate a deep neural network model based on Long Short-Term Memory (LSTM), in comparison to two conventional ML classifiers: Support Vector Machine (SVM) and Multilayer Perceptron (MLP).The authorship classification results indicate that the Quran and Hadith were likely authored by two different Authors, as evidenced by the high classification accuracy: 99% with SVM, 99% with MLP, and 100% with the LSTM-based deep neural network, which outperformed the other ML classifiers. This study aligns with previous research works in the field, and consequently supports the authenticity and originality of the holy Quran.