Interpretable Charge Predictions for Criminal Cases: Learning to Generate Court Views from Fact Descriptions

Proceedings of NAACL-HLT 2018, pages 1854–1864 New Orleans, Louisiana, June 1 - 6, 2018. c⃝2018 Association for Computational Linguistics Interpretable Charge Predictions for Criminal Cases: Learning to Generate Court Views from Fact Descriptions 1Hai Ye∗, 1Xin Jiang∗, 2Zhunchen Luo∗, 1Wenhan Chao† 1 School of Computer Science and Engineering, Beihang University 2 Information Research Center of Military Science PLA Academy of Military Science 1,2 Beijing, China 1 {yehai, xinjiang, chaowenhan}@buaa.edu.cn 2 zhunchenluo@gmail.com Abstract In this paper, we propose to study the prob- lem of COURT VIEW GENeration from the fact description in a criminal case. The task aims to improve the interpretability of charge prediction systems and help auto- matic legal document generation. We for- mulate this task as a text-to-text natural lan- guage generation (NLG) problem. Sequence- to-sequence model has achieved cutting-edge performances in many NLG tasks. However, due to the non-distinctions of fact descrip- tions, it is hard for Seq2Seq model to gen- erate charge-discriminative court views. In this work, we explore charge labels to tackle this issue. We propose a label-conditioned Seq2Seq model with attention for this prob- lem, to decode court views conditioned on encoded charge labels. Experimental results show the effectiveness of our method.1 1 Introduction Previous work has brought up multiple legal as- sistant systems with various functions, such as finding relevant cases given the query (Chen et al., 2013), providing applicable law articles for a given case (Liu and Liao, 2005) and etc., which have substantially improved the working ef- ficiency. As legal assistant systems, charge predic- tion systems aim to determine appropriate charges such as homicide and assault for varied criminal cases by analyzing textual fact descriptions from cases (Luo et al., 2017), but ignore to give out the interpretations for the charge determination. Court view is the written explanation from judges to interprete the charge decision for cer- tain criminal case and is also the core part in a le- gal document, which consists of rationales and a ∗ indicates equal contribution. † Corresponding author. 1Data and codes are available at https://github. com/oceanypt/Court-View-Gen. charge where the charge is supported by the ratio- nales as shown in Fig. 1. In this work, we propose to study the problem of COURT VIEW GENeration from fact descriptions in cases, and we formu- late it as a text-to-text natural language generation (NLG) problem (Gatt and Krahmer, 2017). The input is the fact description in a case and the out- put is the corresponding court view. We only fo- cus on generating rationales because charges can be decided by judges or charge prediction systems by also analyzing the fact descriptions (Luo et al., 2017; Lin et al., 2012). COURT-VIEW-GEN has beneficial functions, in that: (1) improve the inter- pretability of charge prediction systems by gener- ating rationales in court views to support the pre- dicted charges. The justification for charge de- cision is as important as deciding the charge it- self (Hendricks et al., 2016; Lei et al., 2016). (2) benefit the automatic legal document generation as …

Content truncated to 500 words