Internal Control Defect Recognition Model And Empirical Research

When public companies identify and evaluate internal control deficiencies, they rarelyconsider the risk levels and severities of many deficiencies’ combinations, or just evaluatethe deficiencies’ combinations. This paper is to consider how to achieve the evaluation ofcombined deficiencies’ severities, and achieve the judgment of deficiencies’ typesautomatically. First, this paper builds the standard system of internal control deficiencies’judgment, to identify and judge the specific internal control deficiencies’ types which existin the public companies. Then this paper uses BP neural network’s technology to build themodel of internal control deficiencies’ identification which bases on the BP neural network.This model is using to generally analysis the information of internal control deficienciesand the risk superposition effect of deficiencies’ combination, to automatically identify thepublic companies’ types of integral internal control deficiencies.Meanwhile, this paper uses the sample data of the public companies to train the model,and uses another sample data to test the model. The results suggest that when the internalcontrol deficiencies are divided into three categories(material weakness, significantdeficiency, general deficiency), the accurate rate of the model’s deficiencies identificationis73.33%, while the significant deficiency is classified into material weakness, the internalcontrol deficiencies are only divided into material weakness and general deficiency, theaccurate rate of the model is86.67%. And it suggests that when public companies evaluatethe internal control deficiencies, especially evaluate the multiple combined deficiencies,they can use the model to identify the integral internal control deficiencies’ types andpredict material weakness which possible exist, to automatically determine the internalcontrol deficiencies, reduce the subjective evaluation.