Novel Mechanism to Improve Hadith Classifier Performance

Abstract

Muslims believe that the Sunnah of the Prophet Muhammad (SAAW) is the second of the two revealed fundamental sources of Islam, after the Holy Qur'an. Hadith provides a Gold Standard "ground truth" for Artificial Intelligent (AI) knowledge extraction and knowledge representation experiments. In the present study, the extracted Islamic knowledge represented the focal point of the research; three famous books in Hadith science framed the corpus of the study. This study attempted to explore new approach to classify Hadith using a combination of the expert system and data mining techniques to classify Hadith according to its validity degree (Sahih, Hasan, Da'eef and Maudo'), the proposed Hadith Classifier model was built through learning process, Decision Tree (DT) classifier modeling had been represented by the tree structure model, and the attributes of the instances originally were obtained from the source books. Whilst some attributes were indicated as null values, or missing values. A novel mechanism called missing data detector (MDD) was employed to handle these missing data. This mechanism simulated the Isnad verification methods in Hadith science. The results of the research were compared with the resource books, concurrently with the point of view of the experts in the Hadith science. The findings of the research showed that the performance of DT Hadith classifier had significant effect with MDD, the CCR was sharply increased from (50.1502 %) to (97.597%) Furthermore, the favorable obtained results indicated that the DT Modeling is a viable approach to classify Hadith due to the ease of rules induction and results interpretation.