Research On Theory And Approach Of Enterprise Financial Early Warning Based On Big Data

With the development of economic globalization and Internet economy, competition among enterprises becomes more and more intense. Both investors and companies paid more attention to the warning of financial crisis. However, early warning of financial crisis is always a difficult problem in enterprise management. The current financial crisis warning research mainly focused on models about financial indicators. Their limitations were gradually exposed by the paradox of early warning in each economic crisis. The lag and maneuverability of financial indicators seriously affected the credibility of financial crisis warning. Although, some scholars introduced non-financial indicators to improve models, they were difficult to meet all kinds of companies’ requirements due to the difficulty of data collection and one-sided of non-financial indicators selected. Big data thinking and technology offer a new way for the selection of non-financial indicators.In this paper, the theory and method of enterprise financial early warning are researched based on big data. First of all, evolutionary game theory is used to analyze the failure causes of financial crisis warning, and the unreliability of assumption of real and effective financial data was proved, and then an effective path to improve the effect of crisis warning was found. The system dynamics simulation points out that the development of relative information technology is an important direction to improve the effect of financial early warning. Therefore, this paper proposes that regard Internet users as a sensor to enterprise. Based on the relevant online information on the Internet, sensor signals are formed by mixing the analysis of emotion processing and Internet information release frequency statistics together can cover all dimension information about enterprises. Then, this paper uses big data web crawler and technologies of text emotional tendency analysis to structuralize and simplify enterprise information, builds up a comprehensive big data emotion index for listed companies, and combines with the financial indicators to establish a financial risk early warning model which introduced big data index. The analysis results show that this model is better effective. Experiment results offer a new way for theoretical research on financial early-warning, and provide theory support to forecast listed company financial crisis, and also point out that as a presentation of web2.0 online information, the public opinion influences the enterprise crisis each other. According to the view of ‘creation is the best prediction’, this paper studies systematically on the enterprise crisis counseling based on opinion of enterprise. Studies are supported by the National Natural Science Fund project " the research of information security theory and method based on game theory(No: 61272398)" and the Beijing Social Science Fund key project "the research of financial early warning theory and method based on big data(No: 14JGA001)". There are four innovations in this paper as follows:1. The game mechanism research about financial early warning effect. Aim at the difficult of financial crisis warning, this paper discusses the failing causes of financial crisis warning with evolutionary game theory, and demonstrates the unreliability of assumption of real and effective financial data, and also finds an effective path to improve the effect of crisis warning. The system dynamics simulation points out that the development of relative information technology is an important direction to improve the effect of financial early warning.2. The research of the financial early warning model based on big data. Regarding internet users as sensors distribution in Internet for the enterprise and considering related online information, this paper gets information covers all aspects and dimensions of listed enterprises by the analysis of the emotional processing, and combining with the financial indicators, establishes a financial risk early warning model which introduced big data index.3. The design of topic crawler algorithm and semantic analysis algorithm based on big data. This paper designs a topic crawler algorithm that can handle big data based on support vector machine(SVM) and keyword topic, and constructs a semantic dictionary applicable to the financial field, and then designs a text emotional tendency analysis algorithm. Structuralizing and simplifying Internet information related to listed companies and collecting specific sensor signals to support the sensor model, this paper builds up a comprehensive big data emotion index for listed companies, then comparatively analyzes and exams the predictive effect of the model. Test results show that the financial risk early warning model based on big data has better effectiveness.4. The enterprise crisis management according to financial early warning model based on big data. By the model, this paper finds the interactional relationship between enterprise crisis and public opinion. "Creation is the best prediction", an important work in enterprise crisis management is to govern the fluctuant public opinion of enterprise. Considering the groupment of enterprise public opinion producing and developing process, applying cluster dynamics and evolutionary game theory, on the basis of research on group flow process and individual flow process of relevant enterprise network public opinion, this paper builds the guidance model of the enterprise network public opinion. By multiple agent simulation with the model proposed, this paper tries to seek the optimal guidance strategy in relevant constraints. The results of this paper can provide theoretical foundations for enterprise crisis management in view of public opinion in case that the big data guidance model supplies warning signals.