The Design And Implementation Of Personal Tax Declaration Management System Based On Decision Tree Algorithm

In recent years, with the rapid development of database technology, and advancedlanguage appear constantly progress, coupled with the company, enterprises and institutionsfor the attention of the informatization, all kinds of management information system is moreand more perfect at the same time, also brought a lot of information data. For the tax work,are also facing the same problem. Previous tax information system for the tax data processingbasic around the statistics, analysis, query and other functions, can say, tax data contain a lotof valuable information has not been good use.Data mining technology is a new development of the discipline, it can help the tax stafffrom vast amounts of tax data current has never been seriously utilization value, thuseffectively reduce the operation cost of work, provide a more detailed related services, foundin the process of tax related features, help tax officials to take a more effective means ofmanagement. The new ideas and methods will be not seen in the previous work.Tactic schemes, in this paper, based on the J2EE three-tier architecture solution toseparate the business layer and data layer and user layer, middle layer technology using JSPand Java Servlet, JavaBean. Systems network architecture to B/S (browser server mode,convenient personal tax, tax staff and system administrators use function. system connectionOracle database system by JDBC. System uses the JAVA programming language calledWEKA platform module of the decision tree algorithm to realize the core functions of taxassessment system module, and the final evaluation results to the JSP page is displayed on theuser’s browser.In this paper, system function module is divided into tax filing management, basicinformation management, analysis, evaluation management, permissions, distributionmanagement and system management five parts.(1) Declare management module. Including personal basic information input, reportforms to fill out, tax returns, submit and receive feedback information, print, tax paymentreceipts, etc. (2) Basic Information Management include the taxpayers, tax officials andadministrators of individual account management, tax officials for system data query statisticsmanagement functions.(3) Tax Assessment Management is the focus of this article, using the data miningtechnology to realize data information classification of taxpayers, realize to taxpayers taxbehavior, the supervision of any errors found, find out the questionable data.(4) User permissions Management include the system role management, users and usergroups, privilege allocation and so on.(5) System Management manages system server, router, application, parameterconfiguration, queue management and configuration of data exchange platform.One key module for the analysis evaluation, through the comparison of multiple dataclassification algorithms, and analysis of the characteristics of tax data and tax evaluationindex, this paper determined using C4.5decision tree algorithm. C4.5decision tree algorithmis modified based on ID3algorithm, ID3algorithm USES information gain as the attributeselection metric, information gain is defined as the original information demand and thedifference between the new information requirements. By selecting the attribute with thehighest information gain as the node N split attribute to ensure a simple tree construction. Inthis paper, the improvements of C4.5algorithm for discrete attributes, considering the manybranch as the number of nodes in the subtree attributes are equal, and the branch informationuniformity that entropy is not very ideal, too big or too small, so consider incorporating thesebranches, and recount the merged sample information entropy, choose the best test evennessproperties.In this paper, finally, the system function and performance test, function test objectsincluding database links, forms, etc., performance testing, including the evaluation model ofaccuracy and recall rate, production and use of time, for large data sets adaptability of thealgorithm, model analysis to evaluate the comprehensibility of the result. Test results showthat the system model can provide help for the tax staff work less, and be able to help the taxstaff to determine the key object of tax supervision, improve the level of the tax work.