Enhancing LSTM and Fusing Articles of Law for Legal Text Summarization

The growing number of public legal documents has led to an increased demand for automatic summarization. Considering the well-organized structure of legal documents, extractive methods can be an efficient method for text summarization. Generic text summarisation models extract based on textual semantic information, ignoring the important role of topic information and articles of law in legal text summarization. In addition, the LSTM model fails to capture global topic information and suffers from long-distance information loss when dealing with legal texts that belong to long texts. In this paper, we propose a method for summarization extraction in the legal domain, which is based on enhanced LSTM and aggregated legal article information. The enhanced LSTM is an improvement of the LSTM model by fusing text topic vectors and introducing slot storage units. Topic information is applied to interact with sentences. The slot memory unit is applied to model the long-range relationship between sentences. The enhanced LSTM helps to improve the feature extraction of legal texts. The articles of law after being encoded is applied to the sentence classification to improve the performance of the model for summary extraction. We conduct experiments on the Chinese legal text summarization dataset, the experimental results demonstrate that our proposed method outperforms the baseline methods.