Abstract
Women frequently seek guidance on these matters because they directly affect daily religious practices, yet reliable information can be difficult to access quickly. To address this need, the study implements an intelligent chatbot capable of providing accurate and relevant responses. Therefore, The development of an automated question-answering system designed to support learning and understanding of women's Islamic jurisprudence (fiqh), particularly the topics of menstruation, postpartum bleeding, and istihādah. The system utilizes the Indonesian Bidirectional Encoder Representations from Transformers (IndoBERT) model as a natural language processing engine and is integrated into the Telegram platform for ease of user access. This model facilitates the acquisition of information related to women's fiqh through an automated question answering system. There are two IndoBERT models were used in the fine-tuning process, including Rifky/Indobert-QA and Wikidepia/Indobert-lite-squad. The results of this study indicate that the Rifky/IndoBERT-QA model provides better performance in the question-and-answer system for the menstruation, postpartum bleeding, and istihadah domains. Evaluation using BERTScore yielded a 73.7% of precision, 73.4% of recall, and an 73.3%. of f1-score. The user performance evaluation used a Mean Opinion Score (MOS) with an average value of 4,09, indicating that the chatbot was considered good. Overall, this research contributes valuable insights into the application of Transformer-based language models for religious knowledge dissemination and demonstrates the potential of IndoBERT as an effective solution for specialized Islamic jurisprudence domains for women’s need.