Granule description in knowledge granularity and representation

Jinhai Li

doi:10.1016/j.knosys.2020.106160

2020-09-01

PRE

李金海 (Jinhai Li) *

Zhiming Liu

DOI：10.1016/j.knosys.2020.106160

原文链接

原文求助

摘要

En 中文

Granule description is an important problem in knowledge granularity and representation. In the theoretical study of granule description, two basic sub-problems need to be solved: (1) check whether a target granule is definable, and try to offer an exact description for the target granule if definable; (2) an appropriate approximate description should also be provided to the target granule if indefinable. Up to now, there have been some preliminary studies on these two sub-problems. However, the existing findings are not enough or incomplete since sufficient and necessary conditions of definable granules have not been explored, and there have not been effective ways to find better approximate descriptions of indefinable granules. Motivated by these discussions, this paper puts forward granule description methods for definable and indefinable granules. First of all, we propose the notions of covering elements and inserting elements of a target granule by which sufficient and necessary conditions of boolean AND-definable and (boolean AND, (sic))-definable granules are constructed. And then, the smallest definable granule containing the target indefinable granule and the largest definable granule included in the target indefinable granule are investigated, which are used to solve the problem of finding better approximate descriptions for indefinable granules. Finally, a comparison is made to show that the existing work on complete sets should be improved for providing feasible solutions to the granule description problem.

Keyword:

Knowledge granularity

Knowledge representation

Granular computing

Granule description

Definable granule