Existing research on judicial sentencing prediction often lacks interpretability, necessary for scholarly research and judicial practice.
A new Saturated Mechanistic Sentencing (SMS) model is proposed, rooted in China's Criminal Law, to provide legal interpretability.
The Momentum Least Mean Squares (MLMS) adaptive algorithm is introduced for the SMS model, with a focus on accuracy without data assumptions.
Experiments with the Chinese Intentional Bodily Harm dataset show promising prediction accuracy close to the theoretical upper bound, validating the model and algorithm.