电子学报 ›› 2021, Vol. 49 ›› Issue (4): 760-767.DOI: 10.12263/DZXB.20200098

• 学术论文 • 上一篇    下一篇

基于模板张量分解和双向LSTM的司法案件罪名认定

李大鹏1,2,3, 陈剑2,3, 王晨2,3, 闻英友2,3, 赵大哲1   

  1. 1. 东北大学计算机科学与工程学院, 辽宁沈阳 110819;
    2. 东北大学东软研究院, 辽宁沈阳 110819;
    3. 辽宁省工业控制安全技术工程中心, 辽宁沈阳 110819
  • 收稿日期:2020-01-15 修回日期:2020-09-22 出版日期:2021-04-25 发布日期:2021-04-25
  • 通讯作者: 李大鹏
  • 作者简介:陈剑 男,1980年9月出生,东北大学副教授,硕士生导师,主要研究方向为人工智能、网络与信息安全.E-mail:chen.jian@neusoft.com
  • 基金资助:
    国家自然科学基金(No.61972079,No.61772126);国家重点研发计划(No.2018YFC0830601);教育部基本科研业务费(No.171802001,No.2016002,No.2016004);辽宁省重点研发计划(No.2019JH2/10100027);辽宁省"兴辽英才"计划项目(No.XLYC1802100)

Conviction in Judicial Cases Based on Template Tensor Decomposition and Bidirectional LSTM

LI Da-peng1,2,3, CHEN Jian2,3, WANG Chen2,3, WEN Ying-you2,3, ZHAO Da-zhe1   

  1. 1. School of Computer Science and Engineering, Northeastern University, Shenyang, Liaoning 110819, China;
    2. Neusoft Group Research, Northeastern University, Shenyang, Liaoning 110819, China;
    3. Research Center of Safety Engineering Technology in Industrial Control of Liaoning Province, Shenyang, Liaoning 110819, China
  • Received:2020-01-15 Revised:2020-09-22 Online:2021-04-25 Published:2021-04-25

摘要: 案件罪名认定是司法业务的重要环节,尚缺乏有效的智能辅助工具和手段.针对案件定罪的难点问题,提出一种结合张量分解和双向LSTM (Long Short-Term Memory)神经网络的案件定罪方法.该方法将案件数据表示为张量,并在张量分解过程中引入模板张量.模板张量可以在双向LSTM神经网络分类模型的训练过程不断的被优化,使得分解后的核心张量包含更加有效的张量结构和特征信息,有助于提高后续分类模型的准确性,实现案件罪名的精准认定.实验结果表明:所提出的基于张量分解和双向LSTM的司法案件定罪方法比现有方法具有更好的准确性.

关键词: 张量分解, 双向长短期记忆, 模板张量, 案件定罪, 文本分类

Abstract: Conviction in judicial cases is an important part of judicial business,but there is still a lack of effective intelligent auxiliary tools and methods.Aiming at the difficult problem of conviction in judicial cases,a method combining tensor decomposition and Bi-LSTM neural network is proposed.This method represents the case data as a tensor and introduces a template tensor in the tensor decomposition process.The template tensor can be continuously optimized during the training process of Bi-LSTM neural network classification model,so that the decomposed core tensor contains more effective tensor structure and feature information,which is helpful to improve the accuracy of the subsequent classification model and realize the accurate conviction in judicial cases.The experimental results show that the proposed method for conviction in judicial cases based on tensor decomposition and Bi-LSTM has better accuracy than the existing methods.

Key words: tensor decomposition, Bi-LSTM, template tensor, conviction in judicial cases, text classification

中图分类号: