Research On The Method Of Prediction Of Audit Suspects Based On Natural Language Processing Technology Under The Background Of Informationization

The introduction of implementing full audit coverage makes the problems of tight time,heavy task and small quantity of auditors more visible.The efficiency of traditional audit work still needs to be improved.Efficient auditing can not be achieved without the application of information technology in audit business process.In the current government audit practice,auditors neglect the application of information technology in the early stage of audit work.As an important result of the pre-audit work,the forecast of audit doubtful points provides specific direction for carrying out audit verification work.However,the traditional way of artificially predicting audit doubt points reduces the efficiency and effectiveness of audit work to a certain extent due to the large amount of policy reading,the difficulty of quantifying policies and the uneven ability of auditors.Natural language processing technology based on machine learning can effectively simulate the process of auditors predicting audit doubt points,which is suitable for studying prediction of audit doubt points.Therefore,this paper attempts to explore the method of predicting audit doubt based on natural language processing technology,and expects to assist auditors,especially those who lack experience,to carry out audit work efficiently.This paper first introduces the related contents of audit information,audit doubt and natural language processing,analyzes the difficulties and shortcomings faced by the traditional artificial forecast audit doubt,and the necessity of predicting audit doubt based on natural language processing technology in the information background.It provides a theoretical basis for constructing the model of predicting audit doubtful points based on natural language processing technology and case analysis.Then a model of intelligent prediction audit doubt with universal application is constructed by using natural language processing technology such as jieba word segmentation,skip-gram model,smooth inverse frequency,KNN smoothing and so on.Finally,the model is applied to the intelligent forecast of the doubtful points of the employment subsidy fund and the unemployment insurance fund audit project,and a set of audit doubtful points to be verified is obtained.In the process of predicting audit doubt based on natural language processing technology,the computer first refine the audit scheme through intelligent interpretation,and automatically identify and mark the target subject,the main reference object and the main operating object of the audit project,so as to pave the way for the follow-up work.Then,according to the mark,the computer selects and extracts the regulations and policies related to the achievement of this audit goal in the massive policy,so that the system can analyze and learn all the relevant policies,so as to further predict the audit doubt from the legal provisions.Finally,in accordance with the regulations and policies and the audit programmes handled by the computer,combined with previous audit cases,the computer intelligently forecast audit doubtful points,and assists the auditors to carry out the audit evidence collection work.In the scope of documents that can be retrieved at present,no scholars have put forward the use of natural language processing technology to predict audit doubts.The results of this paper show that,compared with the traditional artificial prediction method,based on the natural language processing technology can get the same quantity and quality audit doubt with shorter audit time,which has practical application value to improve the audit efficiency and effect in the audit practice.