Application of Decision Tree ID3 Algorithm in Tax Policy Document Recognition

Abstract

AbstractIn recent years, the tax system has vigorously promoted office automation, and more and more data needs to be processed. This puts forward higher requirements for the classification of massive data and the output of tax knowledge, but the current tax system is still in the simple data search stage. The processing of official documents in the taxation system is the core of office automation and the main method for issuing tax policies. Since the operation of the tax system, a large amount of official document data has been generated, which contains unknown and possibly useful tax information. Taxpayers must keep abreast of and implement the latest tax policies through formal document processing. Therefore, how to quickly and effectively identify tax policies is a major challenge facing current tax information. This article explores and studies the decision tree ID3 algorithm in the recognition of tax policy official documents. It has a general understanding of the text recognition process on the basis of literature data, and then proposes the application advantages of the decision tree ID3 algorithm on the basis of it, which provides the following experiments. The idea is that according to the tax policy document recognition experiment based on the decision tree ID3 algorithm, the actual tax policy samples are 43, and the decision tree ID3 algorithm has identified 41, with an accuracy of 97%. In order to further verify the superiority of the algorithm, on the basis of comparison with the other two algorithms, the experimental results show that the classification accuracy of WNB, SWNB and decision tree ID3 are compared. It is not difficult to find that the classification of official documents in the field. Decision tree ID3 has achieved good classification results, and its performance is significantly better than WNB classification algorithm and SWNB classification algorithm.KeywordsId3 algorithmTax policyDocument recognitionDecision tree