摘要
法律判决预测是基于法律法条规定以及对大量判例的案件信息与法律后果之间关系的科学分析,从而对尚未判决案件法律后果的一种或然性预测预判。本文以中国裁判文书网2015年~2020年的海事裁判文书为数据挖掘对象,通过OCR图像文本识别,并将非结构化数据转换为结构化数据,然后依据CRISP-DM的文本挖掘流程,在运用N-Gram算法去除多余虚词的基础上,运用关键词权重分析法(TF-IDF)和关联性分析法,对海事判决书中的关键词进行分类和数据转换,再通过对案件全流程模块化拆分、关键词触发集合建模的关联分析及匹配结果,针对一定量的训练数据通过相关分析和回归分析来输出预测裁判结果。
The prediction of legal decisions is based on the provisions of legal laws and the scientific analysis of the relationship between the case information and the legal consequences of a large number of precedents,so as to predict the legal consequences of cases that have not yet been decided.This paper takes the maritime case of the judicial document network from 2015 to 2020 as the data mining object,through OCR image text recognition,and unstructured data is converted into structured data.According to CRISP-DM's text mining process,on the basis of using N-Gram to remove redundant function words,keyword weight analysis(TF-IDF)and correlation analysis are used to classify and convert the keywords in the maritime judgment,Then,through the modular split of the whole process of the case,the correlation analysis and matching results of the keyword triggered set modeling,and for a certain amount of training data,the correlation analysis and regression analysis are used to output the judgment result predicted.
作者
甘正男
苏朝阳
徐琪
GAN Zhengnan;SU Chaoyang;XU Qi(CETHK Group Co.,Ltd.,Hangzhou 311100,China;Shanghai Linggang Law Firm,Shanghai 201306,China)
出处
《智能物联技术》
2022年第3期7-11,43,共6页
Technology of Io T& AI
