E-money is a product of the fusion and development of network technology and digital economy, which has the attributes of data, property and money, and the current Criminal Law’s regulatory response to e-money crimes has highlighted the inadequacy. In order to improve the criminal law regulation of e-money crimes, this paper combines the Siamese-BiLSTM model based on the attention mechanism and TextRank algorithm to design a similar case retrieval model for e-money crimes, in order to analyze the loopholes of the criminal law regulation and the reasons for their emergence, and thus to give the path suggestions. Compared with the baseline model Siamese-BiLSTM, the accuracy and F1 value of the improved model in this paper are improved by 5.09% and 5.20% on average, respectively, and the removal of any module leads to a decrease in model performance. This indicates that the improvement of the e-money crime similar case retrieval model based on SBA+TextRank in this paper is better, and it is applicable to the similarity calculation of legal case texts. This paper provides a feasible path for improving the criminal law system for e-money crimes.