<![CDATA[The administration of justice in Religious Courts faces inefficiencies due to manual case classification and unpredictable resolution timelines. This research addresses the gap in systemic judicial management tools by proposing a hybrid machine learning architecture. The objective is to design a systemic information model integrating Deep Learning for case type classification and Ensemble Learning for duration prediction. The method employs CRISP-DM framework combining Long Short-Term Memory (LSTM) for text analysis and XGBoost for temporal regression on 4,461 case records from Bekasi Religious Court. Results indicate a classification accuracy of 93.9% and a Mean Absolute Error (MAE) of 12.4 days for duration prediction. The novelty lies in the systemic integration of dual predictive models into judicial workflows, proposing a digital transformation model for court administration. Conclusions affirm that hybrid AI systems significantly enhance resource planning and administrative efficiency]]>
Copyrights © 2026
