水产品安全信息系统中属性离散化方法研究

摘要

Continuous attributes discretization is an important work for intelligent analysis in an aquatic product safety information system based on rough sets theory,and it is a hot and hard problem in research fields. To deal with it,a new discretization method was proposed based on relative entropy in rough set. The method took the candidate interval classification entropy as the discretization threshold values, and determined the key discretization values through merging the interval values and deleting the redundant values in order to realize the continuous attributes discretization in aquatic product safety information system. Experimental results indicate the algorithm is effective and feasible.%连续属性离散化作为水产品安全信息系统中进行智能化数据处理的一个重要研究内容，已然成为水产品安全信息化研究领域的一个热点和难点。文中利用基于粗糙集理论相对熵的连续属性离散化方法来解决这个问题。此方法选用候选区间的类信息熵作为离散门限值边界，并且通过考察每个属性值的分类能力，合并离散区间，去掉冗余断点，确定关键离散属性值，最终在水产品安全信息系统中实现连续属性离散化。实例分析表明算法是有效可行的。