The integration of legal studies and artificial intelligence has attracted widespread attention, with Legal Judgment Prediction being one of the most typical applications of AI as a tool. However, in judicial practice, obtaining a reasonable verdict requires the identification and organization of criminal circumstance elements. In this paper, we proposed a scalable determination framework of quantitative elements of criminal offenses integrating XAI (XAI-QED). On one hand, we improve the accuracy of circumstance recognition by introducing a hierarchical structure of circumstances and roles, and explore potential factors that may influence judgment outcomes through similar case analysis. On the other hand, we incorporate an explainability module to enhance the credibility of the framework’s prediction results. Finally, simulation experiments validate the effectiveness of the proposed framework in both circumstance recognition and outcome prediction.