APPLICATION OF TEXT MINING TO CLASSIFY THE CORPUS OF AL-HADITH

• Main Author: Ziad Rahmatullah, Muhammad
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/69190
• Institution: Institut Teknologi Bandung

Hadith is the second source of law for Muslims. With so many recorded hadiths, the narrators have classified these hadiths with different compilation methods in each of their books. The book of Sahih Bukhari has the most number of chapters of the 8 other Sunni hadith books. Therefore, there are chapter titles listed in the book of Bukhari hadith but not listed in other hadith books. To find hadiths that match the chapters that are not listed in a book of hadith, we uses one of the techniques in text mining, namely text classification. Several methods that can be used to perform text classification are Naive Bayes (1761), SVM (1992), RNN (1986), CNN (1998), Information Retrieval (1979), and the most recent in the neural network approach is BERT (2018). Bidirectional Encoder Representations from Transformers or BERT is the most up-to-date technology at the time of this research. BERT has a higher speed and accuracy than other neural network methods. In this study, training was conducted using 3 methods, namely SVM, Naive Bayes, and BERT on the Bukhari hadith book dataset. This training aims so that the resulting model can classify input hadith from other hadith books as one of the chapters of the Bukhari hadith book. Thus, each hadith from another hadith book will have a new classification chapter according to the chapter of the Bukhari hadith book and provide new knowledge that a hadith will enter a specific chapter that is not in the original hadith book.