基于模糊知识粒度的混合属性约简算法

现实世界中常常包含着海量的、不完整的、模糊及不精确的数据或对象，使得模糊信息粒化成为近年来研究趋势。利用论域上的模糊等价关系定义了模糊粒度世界的模糊知识粒度，给出了新的属性约简条件和核属性计算方法，以便更好地挖掘出潜在的、有利用价值的信息。针对粗糙集在对连续属性约简的过程中容易造成信息缺失和不能对模糊属性处理的现象，提出了一种基于模糊知识粒度对混合决策系统约简的启发式算法，省去了连续属性离散化过程，减少了计算量，为离散值域和混合值域约简提供了统一的方法。最后通过实例验证了其有效性。

Hybrid attribute reduction algorithm based on fuzzy knowledge granulation

In the real world there are massive, incomplete, vague, and inaccurate data or objects. As a result fuzzy information granulation has become a research trend in recent years. This paper defines a fuzzy knowledge granulation of fuzzy granular world by making use of fuzzy equivalence relation on the universe of discourse. Then it gives new rules of attribute reduction and a method of computing core attributes in order to preferably dig out some potential and valuable information. Because in a rough set model numerical attribute reduction usually brings information loss and fuzzy attributes are not taken into consideration, a heuristic algorithm for the reduction of hybrid decision system based on fuzzy knowledge granulation is proposed to eliminate the discretization process of continuous attributes, reduce the computational complexity and provide a unified approach for normal values and hybrid data. An example shows that the algorithm is effective.