Affiliation:
1. Department of Information Management, National Formosa University, Yunlin 63201, Taiwan
2. School of Computing and Data Science, Wentworth Institute of Technology, Boston, MA 02115-5998, USA
3. Institute of Information Science, Academia Sinica, Taipei 11529, Taiwan
Abstract
This paper employs an order-theoretic framework to explore the intricacies of formal concepts. Initially, we establish a natural correspondence among formal contexts, preorders, and the resulting partially ordered sets (posets). Leveraging this foundation, we provide insightful characterizations of atoms and coatoms within finite concept lattices, drawing upon object intents. Expanding from the induced poset originating from a formal context, we extend these characterizations to discern join-irreducible and meet-irreducible elements within finite concept lattices. Contrary to a longstanding misunderstanding, our analysis reveals that not all object and attribute concepts are irreducible. This revelation challenges the conventional belief that rough approximations, grounded in irreducible concepts, offer sufficient coverage. Motivated by this realization, the paper introduces a novel concept: rough conceptual approximations. Unlike the conventional definition of object equivalence classes in Pawlakian approximation spaces, we redefine them by tapping into the extent of an object concept. Demonstrating their equivalence, we establish that rough conceptual approximations align seamlessly with approximation operators in the generalized approximation space associated with the preorder corresponding to a formal context. To illustrate the practical implications of these theoretical findings, we present concrete examples. Furthermore, we delve into the significance and potential applications of our proposed rough conceptual approximations, shedding light on their utility in real-world scenarios.
Subject
Geometry and Topology,Logic,Mathematical Physics,Algebra and Number Theory,Analysis
Cited by
1 articles.
订阅此论文施引文献
订阅此论文施引文献,注册后可以免费订阅5篇论文的施引文献,订阅后可以查看论文全部施引文献