Research on the Application of Big Data Decision-making Algorithm in Sentencing

The traditional sentencing method mainly relies on the experience and subjective judgment of judges. Affected by factors such as individual cognitive differences and the complexity of cases, it is difficult to meet the demands of judicial practice for the scientificity and accuracy of sentencing. With the rapid development of big data technology, decision-making algorithms have provided a new perspective for sentencing decisions. In this paper, by collecting and organizing the information of a large number of concluded criminal cases and training the model based on deep learning algorithms, it is proved that deep neural networks can significantly improve sentencing prediction. Especially, the accuracy rates of Bi-LSTM-CNN, Bi-LSTM-attention and Han networks are relatively high, reaching 93% respectively. 92% and 75%. It is indicated that the algorithm model performs well in terms of the accuracy and stability of sentencing prediction, and can provide objective data support and reference basis for sentencing. It has a positive promoting effect on improving the judicial system and enhancing the credibility of the judiciary.