Semantic graph knowledge representation of Al-Quran for question answering system

Al-Quran is one of the primary knowledge resources in Islam, containing a vast amount of knowledge in various domains. Most of the current knowledge representation models for Al-Quran are based on the ontological approach, which focuses on extracting concepts in Al-Quran rather than meanings of the Quranic verses. There is a lack of research that focuses on utilizing word dependencies for capturing the meanings of Quranic verses. This research proposes a semantic graph knowledge representation for Al-Quran using word dependencies. The proposed model obtains dependencies between the words in Quranic text through dependency parsing. Based on syntactic and semantic analyses, a set of rules is developed for generating semantic triples representing the meanings of the Quranic verses based on their word dependencies. The semantic triples are mapped into a graph database as semantic dependency graph. The knowledge representation model has been tested in a question answering experiment. The results are evaluated for retrieval accuracy using Precision, Recall and F-score metrics. The proposed model has achieved 62.7%, 53.3%% and 57.7% for the respective metrics. Therefore, the rule-based semantic graph representation model for Al-Quran is a viable approach to represent semantic knowledge of Quranic verses.