Quranic texts retrieval in Indri

Abstract

This paper investigated the retrieval performance of Quranic texts translation of Pimpinan Ar-Rahman in Jawi Malay (the old version of Malay Language) and Rumi Malay (the new Roman version of Malay Language) with different version of Malay Language and letter scripting considerations. The purpose of this study is to test retrieval performance of Quranic text from various scripts of the same source. Methodology implemented was by using queries formulated from Frequently Asked Questions (FAQs) written by Muslim scholars in authenticated printed sources produced in Malaysia. 13 FAQs were selected from Utusan Melayu newspaper for Jawi Malay queries and 13 FAQs from Berita Harian newspaper for Rumi Malay queries. Keyword(s) were derived by researcher from FAQs as queries to represent each FAQs. Matching process was conducted by posting queries into Indri Retrieval. Result from matching process produced results of list of Ayat and these Ayat undergo relevance judgement process that involved three human judges to determine the relevancy of FAQ topic and Ayat, while verifying the accuracy of retrieval. Data analysis that consisted of three measures; precision, recall and overlap discovered competency of Indri for handling multi-script Quranic texts retrieval.