Research On Credit Risk Assessment Of Listed Manufacturing Companies With Text Information

In recent years,with the rapid development of the Internet,news media,analysts and investors like to express their views on the financial market through the Internet,which has generated a large number of text information on the Internet.In the financial market,the news media play the role of information intermediary,disseminating or relaying information related to corporate profits and losses,corporate management and so on.To a certain extent,this information affects the decision-making behavior of investors,thus affecting the performance of enterprises.At present,it is difficult that if we want to extract valuable information from the vast amount of financial news to judge the financial situation of enterprises.Based on this,in order to effectively extract useful information for investors from the vast amount of financial news,this paper proposes an emotional analysis method of financial text based on dependency syntax analysis according to the characteristics of financial news text data structure.The main contents of this paper include the following four aspects.Firstly,by combing the relevant research results of predecessors,this paper analyses the mechanism of text information in credit risk assessment from four perspectives-----the important information dissemination of enterprise value,the warning of enterprise default tendency,the warning of deterioration of enterprise credit status and the influence of text information on the psychology and behavior of investors and consumers.Secondly,the characteristics of network text are briefly introduced from three angles-----company report,news report and online comment.At the same time,the principles and characteristics of three text emotion mining methods based on readability,dictionary and machine learning are compared and analyzed,and the text analysis method suitable for this paper is selected.Thirdly,according to the characteristics of text analysis method and the structural characteristics of financial news text,this paper constructs a model to mine the emotional inclination of financial news.At the same time,this paper introduces the construction of sentiment dictionary in financial field in detail by using Pyhton3.6.5 crawler software.Fourthly,through the construction of BP neural network model and the use of SPSS software,this paper makes an empirical analysis on the risk assessment of manufacturing listed companies by using pure financial index system and mixed index system with emotional index.Empirical results show that the risk identification degree of mixed index system model with news sentiment index is 3 percentage points higher than the pure financial index system model,especially for ST listed companies.Whether from the perspective of pure financial index system modeling or from the perspective of mixed financial index system modeling,with the passage of time,the risk identification degree of the model for listed companies gradually declines.