多粒度形式概念分析的不确定性度量机制与知识粒演化方法

多粒度分析是人类认知事物的重要方法之一，它与形式概念分析相融合产生了多粒度形式概念分析理论，为多粒度数据的知识发现提供了新的研究视角。目前，在该理论中已经初步得到了不同粒层之间概念的关联性和跨粒度知识发现方法，但尚未研究知识粒演化的一般规律与粒层变化时所引起的不确定性的度量方法。为此，本项目系统研究多粒度形式概念分析的不确定性度量机制与知识粒演化规律。具体地，基于不同的多粒度粒化策略，在多粒度形式概念分析中给出信息的不确定性度量方法，以此作为多粒度数据粒度组合优劣的评价准则，设计高效的最优粒度选择算法；通过对三元背景的分解与合成，提出多粒度三元概念分析的知识发现理论，扩充多粒度数据的知识发现方法；在粒度空间的动态变化过程中，研究不同类型的概念与决策规则的演化规律，以便给出知识更新的有效方法。本项目的研究成果将为多粒度数据的知识发现与知识更新问题提供多粒度形式概念分析的有效方法。

Multi-granularity analysis is one of the important methods for human cognition. Combined with formal concept analysis, the theory of multi-granularity formal concept analysis is proposed, which provides a new research perspective for knowledge discovery of multi-granularity data. At present, the correlation of concepts among different granularities and the cross-granularity knowledge discovery method have been obtained preliminarily in this theory, but the general rule of knowledge granularity evolution and the measurement method of uncertainty caused by the change of granularity level have not been studied. Therefore, this project devotes to systematically study the uncertainty measurement mechanism and knowledge granule evolution rule of multi-granularity formal concept analysis. Specifically, based on different multi-granularity granulation strategies, information uncertainty measurement method is given in multi-granularity formal concept analysis. This measurement is used as an evaluation criterion of combined granularity in multi-granularity data and the efficient algorithms of optimal granularity selection are designed; through the decomposition and synthesis of triadic contexts, the knowledge discovery theory of multi-granularity triadic concept analysis is proposed, which extends the knowledge discovery method of multi-granularity data; with the dynamic change of granularity level, the evolution rules of different types of concepts and decision rules are studied in order to give an effective way of knowledge updating. The research results of this project will provide a solution to the problem of knowledge discovery and knowledge updating of multi-granularity data under the framework of multi-granularity formal concept analysis.