Abstract:
This research contributes to the area of Question Answering (QA) to develop a novel QA system for the holy Quran. Q A is an important research area concerned with developing an automated approach that answers questions posed by humans in a natural language. None of the existing QA systems on the Quran has been tested on a Quran question and answer corpus using question redundancy with the aim of question answering. In this thesis, a basic beginning arises from the usability and limitation of a restricted domain knowledge base. We considered the special domain knowledge base important and useful to answer new question asked by a user, as it contains a lot of similar or relevant question information. This work aims to compile a question and answer bilingual corpus in order to develop a QA system that answer Arabic and Englishquestionsabout the holy Quran. This corpus has been collected from several credible sources,and hence it canbe used for the evaluation of question answering systems or any other application where questions and answers are needed. Considering these, first we investigated different techniques and tools necessary to build a novel QA system. WEKA and Nooj have been tested and explored by conducting some experiment, but they are not suited to build a QA system. Then we built our own implementation: QAEQAS a Quranic Arabic/English Question Answering System. As a complete QA solution, we used python and its toolkit to process the user question and the corpus, as well as to implement the search engine to retrieve candidate results and then extract the best answer. A central argument of this thesis is that it relies on a specialized search corpus, and data redundancy by integrating a range of knowledge bases from different sources as well as reformulating the corpus questions in different ways in differing context. Two prototype versions of QA system were developed. QAEQAS presents its usefulness that deals with a wide range of question types in addition to its high accuracy. QAEQAS used the most common evaluation metrics in Information Retrieval (IR) and QA toevaluate the effectiveness and correctness of the system results, namely precision and recall. The resultshave shown that the performance of the system is increased with more data redundancy;it registered a precisionand recall of 96% and 94%for Arabic, 90% and 89%for English respectively.
