面向行政处罚的智能决策系统开发及研究

Research and Development of Intelligent Decision System for Administrative Penalty

在行政法律领域中,人工智能与法律有着天然适应性。针对行政执法人员决策处罚不公正、证据保留不充分、行政处罚过重等问题,将决策支持系统应用到行政执法过程中,辅助处罚决策,以便对行政案件进行快速、公正的处理,避免二次事件的发生。考虑智能决策系统的处罚正确性提出基于模糊神经网络和案例推理的专家分析系统,集行政执法的高效率与公平性于一体。面向行政处罚的智能决策系统引人了径向基函数(RBF)神经网络和高木关野(IS)模糊模型进行系统构建。基于模糊推理模块与模型训练方式,对执法的输人量进行处理;根据输出量指导案件的处理,结合处罚相关的专家知识,保证输出方式符合实际的决策意图。通过模拟测试和分析,试验结果验证了智能决策支持系统的高效性与正确性,阐述了违法情节与违法危害程度对处罚结果满足正相关的关系,具有行政处罚参考意义。开发的面向行政处罚的智能决策系统将成为行政裁量的发展方向。

In the field of administration and law,artificial intelligence and law are naturally fitted one another.In view of the decision of injustice penalty,insufficiency of evidence retention and undue administrative penalty by the administration enforcement stuffs,the administrative cases are quickly,justly managed to avoid occurs of secondary events during decision support system adopting administrative enforcement law.In light of penalty correctness of intelligent decision system,an expert analysis system is proposed based on fuzzy neural network and case-based reasoning to integrate administrative enforcement law and fairness in one.Radial basis function(RBF)neural network and Takegi-Sugeno(TS)fuzzy model are built up to orient towards intelligent decision system of administrative penalty.Based on fuzzy ratiocination module and pattern training mode,according to the output of case disposes combining expert knowledge,the output mode is conformed to the practical decision intention dealing with the input of enforcement.Through the mock test and analysis,the exam results show that the efficiency and correctness of intelligent decision support system are proved.The relationship between illegal plot,extent of harm and the penalty result is shown,which is in positive correlation.