An Empirical Study of Law Articles Prediction on Transportation Legal Cases

摘要

Legal intelligence is a prospective field experiencing rapid development in recent years, and research on judicial documents analysis has become a significant and core task in this field. Among these researches, law articles prediction is undergoing slow development and far from being put into practical use. It is actually a multi-label classification task. Previous works mostly regard law articles as merely labels, ignoring the semantic meaning of them. There are also works combining facts with law articles text, treating law articles in the same way as any other textual form. In this work, we take the semantics of law articles into consideration and conduct extensive experiments on transportation legal cases with 9 models including both traditional machine learning approaches and deep learning methods. To better measure, we propose a novel metrics which is appropriate for this task. The result shows the effectiveness of utilizing semantic meaning of legal case.