水产品安全评估体系中属性约简算法研究

Research on Attribute Reduction Algorithm in Aquatic Product Safety Assessment System

针对水产品安全评估体系中评估指标的复杂和冗余问题,提出一种混合信息量的属性约简算法。该算法采用正域和信息量相融合的方法,定义属性不重要性公式。为减少算法的搜索空间,只计算除核属性以外的样例属性不重要性。实例验证和分析结果表明,与启发信息为属性依赖性或属性频率的约简算法相比,该算法具有更好的属性约简效果,可以应用于水产品安全评估体系中评估指标的约简。

Aiming at the complexity and redundancy of the evaluation index in the aquatic product safety assessment system,a newattribute reduction algorithm based on attribute unimportance is proposed. The algorithm uses the method of fusing the positive region and the information,defines the formula of attribute unimportance,and calculates the attribute unimportance of the samples which excepts the core attribute in order to reduce the search space. Experimental validation and analysis showthat the attribute reduction effect of the algorithm is better than that of heuristic information attribute reduction. It can be used to reduce the assessment index in the aquatic product safety assessment system.