Optimizing Quranic Literacy with the Tamam Method: Leveraging Artificial Intelligence for Handwritten Arabic Recognition

Abstract

Indonesia, boasting the world's largest Muslim population, has witnessed a swift augmentation in its Muslim demographic. As of 2020, Muslims in Indonesia numbered 209 million, which surged to 219 million in 2021. Such an observation is alarming, especially given the Quran's centrality in Islamic teachings and the profound link between grasping its tenets and the capability to read and write its verses. This paper introduces an innovative application employing the Tamam method, optimized for enhancing Quranic literacy through the recognition of handwritten Arabic texts using Convolutional Neural Networks (CNN). Involving a cohort of 144 participants, who answered 65 questions, a dataset encompassing 3,842 data points was curated for testing and validation. Preliminary results showcased the model's evolution, with a notable rise in accuracy from 14.27% in the initial epoch to 88.87% in the 20th epoch. Despite such advancements, fluctuations in the validation data hinted at potential overfitting scenarios. This study demonstrates the feasibility of integrating the Tamam method with AI-based handwritten Arabic recognition as a supportive tool for Quranic writing practice. It paves the way for more resilient and adaptive Quranic educational tools, ensuring learners grasp the Holy Text in its true essence.