基于多粒度的多源数据知识获取

Knowledge acquisition of multi-source data based on multigranularity

多粒度认知能力是人类分析复杂数据的一种常用策略。作为复杂数据类型之一的多源数据,因其数据源头多而使得数据分析变得复杂。受多粒度思想的启发,以多源信息系统为数据基础,基于悲观的决策策略,提出了多源划分约简集的定义。讨论了多源划分约简集与划分约简集之间的关系,并给出了相应的属性特征的判别方法。最后,针对多源决策信息系统,基于乐观的决策策略,提出了多源决策规则。借鉴多粒度模型,从一个新角度所提出的多源数据分析方式进一步丰富了知识获取的方法。

Multigranularity cognition is the common strategy for analyzing complex data.Multi-source data is one type of the complex data,and its knowledge acquisition become more complicated because of its multisource.Inspired by the idea of multigranularity,the multi-source attribute reduction is defined based on the pessimistic decision-making strategy in multi-source information systems.The relationships between the multi-source attribute reduction and the attribute reduction are discussed in detail,and the corresponding judgment method of attribute characteristics are given.Finally,the definition of multi-source decision rule is proposed based on the optimistic decision-making strategy in multi-source decision information systems.On the basis of multi-granularity model,the proposed method gives a new perspective of multi-source data analysis,which enriches the study of knowledge acquisition.