Research On The Method Of Element Identification And Element Relationship Extraction In Criminal Cases For The News Involved

Automatically extracting case elements and the relationship between case elements from the news texts of criminal cases is an important basis for analyzing the news texts involved and understanding the semantics of the news texts,as well as the basic research work for mastering the public opinion related to the case.This paper mainly focuses on the identification of case elements and the extraction of case element relations in criminal case news texts.The first is to automatically obtain the criminal case news text from the Internet page.Then automatically identify the elements of the case from the criminal sentences of the criminal case,and further automatically extract the semantic relationship between the elements of the case.Finally,the relationship between the extracted case elements is drawn into a relationship network for visual display.The main research work includes:(1)Aiming at the identification of criminal case elements in the news texts involved in the case,a method of criminal case element recognition based on the BERT pre-training model is proposed.First,based on the pre-trained BERT model,the semantic representation of the news text is improved to improve the ability of semantic representation.Then the context semantic information is extracted based on the bidirectional GRU,and the high-level semantic information is extracted based on the attention mechanism.Finally,the conditional random field model is integrated to realize the identification of case elements.The experimental results show that the proposed method can effectively improve the ability to identify the elements of the news texts involved in the case and achieve accurate identification of the elements of the case.(2)Aiming at the problem element extraction of case-related news texts,a model of element relationship extraction based on the background information of the case is proposed to realize the extraction of element relationships guided by case information.First of all,semantic representation of news text and case description information is carried out based on the bidirectional GRU and CNN respectively.Then the attention mechanism is introduced to fuse the case description information with news sentences to obtain the semantic representation of the sentence that fuse the case description information.Finally,through the Softmax classification method to achieve the case element relationship extraction.The experimental results show that this method can improve the effect of extracting the relationship of elements in criminal cases.(3)Network visualization of criminal case elements based on graph database neo4 j.First,collect news data on criminal cases involved;Then,identify the semantic relationship between case elements and extract case elements;Finally,the extracted case element relationships are imported into the graph database neo4 j,and a criminal case element relationship network is drawn.