Font Size: a A A

Research On Multi-class Standard Text Classification Algorithm For Identifying Key Legal Factors Of Judicial Judgment Documents

Posted on:2020-10-23Degree:MasterType:Thesis
Country:ChinaCandidate:S H XuFull Text:PDF
GTID:2506306200450894Subject:Software engineering
Abstract/Summary:PDF Full Text Request
In recent years,the state has vigorously promoted the construction of information technology in accordance with the law.The results of judicial trials have been open and transparent,and a large number of judicial decisions have been made public through the Supreme Court’s network of adjudication documents.In the new round of judicial reform,the judge responsibility system puts forward higher requirements for the trial consistency of judicial cases.In order to ensure the uniformity of the standards of the case and the uniform application of the law,it is necessary to establish a mandatory search mechanism for class cases and related cases.At present,the work of recommending the referee documents in the open literature is very limited.The research on the case retrieval is based on the vectorized representation of the entire legal text content,and the cosine similarity of the vector is calculated to retrieve the case.The characteristics of knowledge,the effect is not good.The algorithm studied in this paper simulates the behavior patterns of the judicial personnel to read the key factors of the case identification law,and retrieves more accurate cases by identifying the key legal viewpoints that determine the similarity of the cases from the original content of the judgment documents.The method studied in this paper has some innovations in the semantic understanding of the legal viewpoints of judicial cases.It provides a new solution to the study of the mandatory retrieval of judicial cases through the classification of legal key factor classification algorithms.This paper solves the problems faced by the multi-class text classification task of the refereeing documents from the following three aspects.Firstly,a SVM combined multiclassification algorithm framework based on sliding window sequence reading is studied to realize the identification of different legal opinions in the original document.Secondly,a CBOW-based class-sequence optimization algorithm is studied to improve the performance of the multi-class identification.Through a large number of experiments on the key factors of the legal identification of the civil lending cases,the algorithm studied in this paper can effectively identify the legal key factors contained in the case and the original claim.Finally,based on the trained model,design and implement a class-based retrieval system based on legal key factors,which can analyze the case of the user’s query and visually display the search results,helping the user to quickly understand the case and the outcome of the case.
Keywords/Search Tags:support vector machine, text classification, word2vec word vector, sequential reading, CBOW, Multi-labels classification
PDF Full Text Request
Related items