Multi-Label Topic Classification of Hadith of Bukhari (Indonesian Language Translation)Using Information Gain and Backpropagation Neural Network

Hadith is the second source of law and guidance for Muslims after the Qur'an; it is a collection of traditions containing sayings of the Prophet Muhammad, with accounts of his daily practice (the Sunnah). Experts of hadith have narrated numerous ahadith. This research builds a system that can classify hadith saheeh (sound)compiled by Bukhari that was translated to the Indonesian language. This area of study is important, so that Muslims could easily recognize the suggestions and restrictions contained in a hadith. The Backpropagation Neural Network algorithm was chosen because it can perform classification with a huge number of varied features. This method is combined with information gain as feature selection in order to select influential features for each class label of multi-label data and single-label data, which has never been done before. The results show that 88.42% of multi-label hadith data can be correctly classified. Meanwhile, 65.275% of the single-label hadith data can be classified correctly using the information gain feature selection.