Dynamic Updating of Association Rules for Multi-granularity Formal Context

Abstract

Abstract Based on granularity tree, this paper studies the changes of association rules before and after attribute granularity transformation in formal context, and proposes zoom algorithms to update association rules. The zoom-in algorithm is to change the granularity of the attribute from coarse to fine, and the zoom-out algorithm is to change the granularity of the attribute from fine to coarse. According to the zoom algorithms, it can directly operate on the association rules in the original formal context by using concepts as a bridge to generate association rules in the new context. Which can reduce the reconstruction process of the concept lattice when the granularity of attributes changes. The experimental results show that the algorithm proposed in this paper greatly reduces the computational workload, and shortens the running time than the classical algorithm flow.