Research On The Application Of Entity Relation Extraction In Financial Credit Investigation

With the rapid development of Internet finance,enterprises and individuals borrow more and more frequently.Banks and other financial enterprises with borrowing as the main business,in order to reduce the risk of capital loss,not only examine the credit and risk situation of the loan entity itself,but also examine the credit and risk situation of the entities associated with the loan entity.In the past,because of the lack of information,financial enterprises often through physical investigation to find the relationship between entities.With the development of the Internet,more and more financial firms want to automatically extract the relationships between entities from textual dataThe main content of this paper is the application of entity relation extraction in the field of financial credit.The main work as follows:1)For the lack of the entity relation category system which conforms to the financial credit business,this paper summarizes a set of entity relation category system that conforms to the financial credit business from the Freebase knowledge base,the referee instruments and the financial class website;2)For the lack of labeled dataset issues,this paper retrieves the data of the referee documents from the Chinese referee documents website and constructs about 10,000 labeled datasets by using the method of artificial labeling and distant supervised learning;3)This paper presents a method of entity relation extraction based on single sentence which take sentence level feature?lexical level feature and entity tag feature as classification features.4)In order to make full use of the document context information in the process of entity relation extraction,this paper proposes a document-based entity relation extraction method,which can extract the relationship from the perspective of the whole document by means of constraint satisfaction problem solving.