面向属性概念的属性粒构建

Attribute granule construction for attribute-oriented concept

通过研究面向对象概念的粒描述,在现有的基于必然属性分析视角的粒描述基础上,讨论了基于粒计算和面向属性概念的属性粒描述。首先,对属性粒进一步拓展,定义了一元可定义属性粒、一元属性描述子。其次,定义了二元可定义属性粒、二元描述子以及基于面向属性概念的M-三支概念,并给出了M-三支概念格的构造算法。最后,通过对三种近似描述的精度比较,表明了基于二元属性描述子的近似描述方法优于基于一元属性描述子的近似描述方法,验证了该描述方法的可行性、有效性。

Through the granule description for the object-oriented concept, based on the granular computing and attribute-oriented concept, the attribute granule description is discussed. At the beginning, unary definable attribute granule and unary attribute descriptors are defined with further expand of the attribute granule. After that, binary definable attribute granule, binary attribute descriptors and M-three-way concepts based on attribute-oriented concept are established. Then, the construction algorithm of M-three-way concept lattice is proposed. Finally, three types of approximate description methods are compared in terms of description accuracy. The results show that the approximate description method based on the binary attribute descriptors is superior to that based on the unary attribute descriptors. The experiment results verify the feasibility and effectiveness of the description method.