With the rapid development of emerging technologies, legal text datatization provides the possibility of intelligent legal judgment assisted decision making. In this paper, a multi-task judgment prediction model based on data mining algorithm and Lawformer is designed to introduce Lawformer pre-training model, incorporate legal theories, and utilize HAN encoder to encode the case text at the word level and sentence level in order to better capture the semantic information. Then the support vector machine algorithm is integrated into the model to explore the crime composition, and a multi-task learning framework is constructed for the intelligent legal judgment assisted decision-making model. The experimental results show that the model in this paper performs well on the CAIL2018-Small public dataset, which significantly improves the model’s legal judgment prediction accuracy and interpretability, with micro-averaged F1 values of 87.3 and 85.4 on the two tasks of crime prediction and law recommendation, and the overall legal case prediction correctness is as high as 99.48%. The research in this paper provides a new working path for the field of intelligent justice, which contributes to the scientific and intelligent development of legal judgment assisted decision-making.
