Classification of Hadith Quality Based on Matan Using LSTM

Hadiths have an important role for Muslims. Hadith is one of the main sources of law for Muslims after the Quran. A hadith consists of two important parts, the sanad and the matan. The sanad is the series of narrators who convey the hadith while the matan is the content of the hadith. Considering the large number of traditions and the abundance of hadith data, the use of Machine Learning or Deep Learning is very appropriate. Therefore, in this study, a system will be built to classify the quality of hadith based on their matan. The classification of hadith quality is not to replace the role of scholars, but rather to help scholars determine the quality of hadith given the large number of hadith. To classify the hadith quality based on its matan, one of the deep learning methods, namely Long Short-Term Memory (LSTM) will be used in this study. LSTM was chosen because it has better performance when compared to traditional machine learning, Recurrent Neural Network (RNN), and Convolutional Neural Network (CNN) for text classification. This research will classify hadith into three classes, namely sahih, hasan, and daif. Before the model training process, sentence embeddings will be used to represent the vector of a sentence. The test results show that the performance of LSTM is superior to RNN, CNN, Decision Tree, and KNN. The best model from this study obtained an accuracy of 0.9605.