SEARCHING AL-QURAN VERSES WITH HADITH AS QUERY USING INFORMATION RETRIEVAL

• Main Author: Rahmadi Munly, Harry
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/63919
• Institution: Institut Teknologi Bandung

Al-Quran and hadith are the main religion guidance for Muslims. Many attempts to interpret these two entities have been done before by Islamic cleric in book form. A collection of verses related to a certain hadith also has been interpreted and found their linkages. However, efforts to digitize the search for the digital Al-Quran through digital hadiths have not been carried out. For this reason, in this final project, a search system is designed using information retrieval in the form of an android application that can search for verses that are relevant to a hadith. The information retrieval built in this final project has variations in the term weighting and document similarity modules. The modules are weighted TF-IDF cosine similarity algorithm, without term weighting jaccard similarity algorithm, word embedding fastText word mover's distance (WMD) algorithm. These three pairs of modules were implemented and their performance compared. WMD also compares the Wikipedia model with the general context domain and Al-Quran Hadith model with the Islamic context domain. Based on the evaluation of information retrieval performance using micro-average precision, cosine similarity with TF-IDF weighting has the highest precision value of 55.49% than other methods being the most suitable method for searching Al-Quran verses using hadith. Based on the performance evaluation, it was also found that the Al-Quran Hadith model obtained precision 33.50%, 1.25% higher than the Wikipedia model, proving that the fastText model should be trained with a corpus that matches its context domain with the document being searched through information retrieval which in this final project has the Islamic context domain.