Rough Approximations and the Minimal Set in Knowledge Spaces

duoqian miao,yiyu yao,lai wei,feifei xu

DOI: https://doi.org/10.1166/jama.2013.1040

2013-01-01

Journal of Advanced Mathematics and Applications

Abstract:The theory of knowledge spaces represents a new paradigm in mathematical psychology for knowledge assessment. As a sub theory of granular computing, it shares many similar properties with rough set theory. In this paper, we first generalize rough approximations. To combine the two sub theories of granular computing, the rough approximations in rough sets are introduced into knowledge spaces. The lower approximation in knowledge spaces represents the question sets which the student certainly master, while the upper approximation in knowledge spaces means the question sets which the student can possibly answer correctly. Furtherly, we interpret the disjunctive model of skill map in knowledge spaces as an or-binary relation table between skills and questions. There may exist skills that are the union of other skills. Omitting these skills will not change the knowledge structure, but can describe the relation between questions and skills much more concise and clearly. Finding a minimal skill set may be formulated similar to the problem of attribute reduction in rough set theory, where an and-binary relation table is used. In this paper, an or-relation skill-question table is considered for a disjunctive model of knowledge spaces. A minimal skill set is defined and an algorithm for finding the minimal skill set is proposed. An example is used to illustrate the basic idea. By introducing the main concepts "rough approximations" and "reduction" in rough sets into knowledge spaces, we can get more insight into the two theories and better comparison, as well as the deep understanding of granular computing.