Transaction information identification method and system based on graph neural network, and medium

Abstract

The invention discloses a transaction information identification method and system based on a graph neural network, and a medium. The method comprises the steps of obtaining a to-be-identified text; performing feature extraction and label prediction on the to-be-recognized text to obtain a labeling result of entity elements in the to-be-recognized text; constructing a corresponding entity relation graph according to the labeling result of the entity elements; performing feature learning on the entity relation graph through a graph attention network and then outputting entity node feature vectors; and performing feature multi-classification on the entity node feature vector, and outputting a transaction element category of each entity node. According to the method, the relation between the entity elements is subjected to feature learning and classification by constructing the entity relation graph, the transaction mechanism category of each entity is recognized, classification judgment can be more accurately carried out by combining the text features and the relation features of the entity elements, and the efficiency and accuracy of information classification and recognition in the current coupon transaction are effectively improved.

Description

technical field [0001] The present invention relates to the field of computer technology, in particular to a transaction information identification method, system and medium based on a graph neural network. Background technique [0002] Spot bond trading is the most active form of trading in the financial trading market. In spot bond trading, there is a large amount of unstructured transaction dialogue text information, which needs to be sorted into standard structured order information. For example, the dialogue text of the spot bond transaction is "3Y 1000001.IB 3.5% 2kw+1XX institution YY fund to ZZ institution", which means "100001.IB bond, with a term of 3 years remaining, at an interest rate of 3.5%, 2,000 10,000 transaction volume", the transaction method is from "XX institution's YY fund account" to "ZZ institution". [0003] When the transaction volume is large, traders will face huge work intensity, and the text structuring task of spot bond transactions is to au...

AI Technical Summary

Problems solved by technology

[0004] Existing recognition technologies basically use natural language processing technology based on deep learning models to understand and recognize text, and then use logical indexing to extract the expression structure of each text to make separate

Images