期刊名称：Journal of Theoretical and Applied Information Technology
印刷版ISSN：1992-8645
电子版ISSN：1817-3195
出版年度：2016
卷号：83
期号：1
出版社：Journal of Theoretical and Applied
摘要：Automatic Question Answering (QA) is the process of identifying exact answer in response of question asked by the user. Several QA approaches have been proposed for many languages. One of such languages is Arabic which spoken by nearly 200 million people. Approximately 90% of Arab people are Muslims. Many Muslims need religious consultations toward specific issues. Such consultations have to be acquired from a legitimate source such as Islamic scholars. The response toward these consolations is called a Fatwa, which is considered as an Islamic point of view toward such issues. Therefore, there is a vital demand to present a QA system that has the ability to answer Muslims questions based on a collection of Fatwas. Several research efforts have been implemented to provide QA for Islamic domain. However, such efforts have used simple lists of lexicons as a source of knowledge, which leads to several limitations in terms of the correctness of the answers. This is due to the process of analyzing question would not be guided with a rich source of knowledge. Hence, this study aims to propose an ontology-based question answering approach for the domain of Islamic Fatwa. The ontology has been constructed using a collection of Fatwas, which has been collected from Ibn Uthaymeen-Prayer Fatwas. Several pre-processing tasks have been applied in order to eliminate the irrelevant data (e.g. numbers, non-Arabic letters and punctuation). Furthermore, Term Frequency-Inverse Document Frequency (TFIDF) has been used in order to provide the main concepts of the domain for ontology construction. The proposed ontology-based QA has been evaluated by comparing it with QA approach that uses simple semantic lexicon as a source of knowledge. The proposed method has outperformed the other approach by obtaining an approximately 90% of F-measure. This demonstrates that the use of ontology as a source of knowledge for QA would enrich the process of extracting the answers.
