基于支持向量机和领域知识的纳税评估预警模型

纳税评估是税务稽查的基础和前提,运用科学、合理的评估手段,降低评估风险和成本,提高纳税评估工作的整体效率和效果,成为纳税评估工作的关键问题.在总结纳税评估基本原理的基础上,针对企业的纳税评估问题构建了评估指标体系,提出了将支持向量机(SVM)和领域知识结合建立纳税评估预警模型的方法,首先利用SVM筛选出有涉税疑点的企业,然后在领域知识的指导下结合统计分析方法,进一步找出可疑对象偏离正常统计值的指标,解决了预警模型无疑点指向性的问题,实证分析表明该方法具有较高的可行性.

Tax Assessment Model Base on SVM and Domain Knowledge

Tax assessment is the basis and prerequisite of tax inspection,building evaluation index system on the basic of summarizing the tax assessment principles,and propose a tax assessment model basic on support vector machine and domain knowledge.First,use SVM screening the tax-related doubts of the enterprise,and then combined with statistical methods under the guidance of domain knowledge,solves the problem of undoubtedly point directivity of the model.Validate analysis shows that the model is feasible.