基于GATE的司法案件信息抽取方法研究

发布时间：2019-05-03 18:00

【摘要】：随着我国司法数据公开的推进,越来越多的司法案件判决信息以网页的形式发布出来供公众查询使用。截止2015年3月,全国各级法院公开的案件判决信息已达700万份以上。如果在商业应用上利用这些公开的案件裁判文书信息,就需要对案件的各个核心字段进行结构化处理,这通常由人工操作来完成。面对如此庞大的案件信息数量,人工处理方式显然在效率上显得非常低下。如何运用信息技术简化这一过程便成为当下急待解决的问题。针对这一研究背景,本文提出了基于GATE的司法案件信息抽取方法研究,即采用基于JAPE规则的信息抽取技术,以GATE系统作为论文研究的技术基础和开发平台。论文首先针对与研究命题相关的国内外研究现状进行研究;接着对相关基础理论进行概述,包括GATE系统介绍、JAPE规则基本语法、执行原理和应用情况的介绍。然后,基于GATE系统提供的平台组件设计,重点研究了基于GATE的司法案件信息抽取方法,包括信息的预处理过程结束,各案件信息字段的JAPE规则抽取方法介绍,结果的组织输出等。最后,将论文所研究的信息抽取方法应用于实际批量的裁判文书数据,语料评测结果显示,该方法的抽取准确率可以达到94%以上,召回率可以达到86%以上,整体性能F值可以达到92%以上,证明该方法的科学性和有效性。
[Abstract]:With the promotion of judicial data disclosure in China, more and more judicial case decision information is published in the form of web pages for public inquiry. As of March 2015, the national courts at all levels of the case judgment information has reached more than 7 million. If we make use of the information of the public case adjudication documents in commercial applications, we need to deal with the core fields of the case structurally, which is usually accomplished by manual operation. In the face of such a large amount of case information, manual processing is obviously very inefficient. How to use information technology to simplify this process has become an urgent problem to be solved. In view of this research background, this paper proposes a method of judicial case information extraction based on GATE, that is, adopting information extraction technology based on JAPE rules and taking GATE system as the technical basis and development platform of this paper. Firstly, this paper studies the domestic and foreign research situation related to the research proposition, and then summarizes the related basic theories, including the introduction of GATE system, the basic grammar of JAPE rules, the principle of execution and the application. Then, based on the platform component design provided by GATE system, this paper focuses on the method of judicial case information extraction based on GATE, including the end of the pre-processing process of information, and the introduction of JAPE rule extraction method of each case information field. Organizational output of the result, etc. Finally, the information extraction method studied in this paper is applied to the actual batch of adjudication document data. The results of corpus evaluation show that the extraction accuracy and recall rate of this method can reach over 94% and 86% respectively, and the result of corpus evaluation shows that the extraction accuracy and recall rate of this method can reach 94% and 86% respectively. The overall performance F value can be more than 92%, which proves that the method is scientific and effective.
【学位授予单位】：天津大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：TP391.1

【相似文献】