Abstract: With the rapid development of Internet technology and artificial intelligence technology, how to introduce artificial intelligence into the judicial field has been closely watched by major research institutions. In the traditional trial-assisted work, the main task is to establish the law, the sentence, and the establishment of the crime for the independent case characteristics. However, in this case, the order of case characteristics is easily ignored, especially when it comes to the time series of case characteristics, and the analysis of the sequence of actions, the “sequence problem” may affect the final trial results, such as cases with the same characteristics may lead to the trial result different. This paper studies how to effectively use this information to improve the performance of legal provisions, to improve the performance in the judicial field, and to assist judges and lawyers to make legal decisions more efficiently. By using the factual description of the case and the sequence of criminal acts to predict the relevant laws involved in the case, this paper verifies the validity of the legal provisions based on the sequence of criminal acts.

Key words: judicial field, criminal behavior sequence, text representation, text classification, relevant article prediction

摘要： 随着互联网技术和人工智能技术的飞速发展，如何将人工智能引入到司法领域得到了各大研究机构的密切关注。在传统的审判辅助工作中，主要依托的是对于独立的案情特征进行法条、量刑、罪名的确立。然而，这种情况下，案情特征的顺序容易被忽略，尤其是涉及到案件特征的时间序列、行为序列的分析时，“顺序问题”可能影响最终的审判结果，如相同特征的案件可能导致审判结果的不同。对如何有效地利用这些信息提升法律条文预测的性能进行研究，用于改进在司法领域的表现，并辅助法官和律师等更加高效地进行法律判决。通过利用案情的事实描述和犯罪行为序列来预测案件涉及的相关法条，验证基于犯罪行为序列的法律条文预测的有效性。

关键词: 司法领域, 犯罪行为序列, 文本表示, 文本分类, 法律条文预测