Islamic Question Answering Systems Survey and Evaluation Criteria

Abstract

Many researchers built Islamic question-answering systems which find the answer to the question from the Quran, Hadith, or Fatwa text using only efficient retrieval techniques. However, it is challenging to answer all kinds of questions due to the current shortcomings in natural language processing tools. Therefore, in this paper, we review the Islamic question-answering systems that can answer all kinds of questions by building a questions and answers corpus and then using the retrieval technique or pre-training model to answer the user's question. After that, we use thirteen evaluation criteria, such as the search approaches and the system scope, to evaluate these systems. We can conclude from this survey that there are flaws in the existing systems, such as all these systems being unavailable and can answer a limited number of questions.