<![CDATA[This systematic literature review aims to comprehensively analyze Named Entity Recognition (NER) applications in Islamic texts, particularly the Quran and Hadith, across Arabic, Indonesian, English, and Malay languages. Materials comprised studies from major academic databases (2017-2024) implementing various NER approaches on Islamic textual datasets. The majority of studies reviewed focused on Hadith texts, with fewer examining Quranic texts and general Islamic literature. The methodology employed a PRISMA-based systematic review examining architectural components, diverse methodologies, comparative model performance, and extraction challenges in Islamic discourse. Traditional rule-based and statistical machine learning methods remain relevant, particularly in hybrid frameworks. However, the analysis reveals that transformer-based deep learning models consistently achieve superior performance, with the highest F1 Scores. Hadith datasets showed better NER performance than Quranic texts due to Hadith's structured and repetitive nature versus the Quran's greater linguistic diversity and complex syntactic structures. Most studies employed lexical and linguistic features to address distinctive characteristics of religious texts, with significant progress in handling specialized Islamic concepts and multilingual considerations. Despite these advancements, significant challenges persist, including the linguistic complexity of Classical Arabic, the scarcity of high-quality annotated corpora, and the difficulties of domain-specific entity identification. This review provides comprehensive guidance for researchers developing Islamic NER systems by identifying optimal methodological approaches and highlighting performance benchmarks across different text types, thereby enabling the development of more effective, culturally aware NLP systems for Islamic content.]]>
Copyrights © 2025
