集成FCM与信息融合的粗粒度级信息的挖掘算法

Integrated FCM and information fusion algorithm of mining coarsegrained information

如何获取粗粒度级信息是信息管理与信息系统中的研究热点之一。提出一种基于模糊认知图(Fuzzy Cognitive Map,FCM)与信息融合集成挖掘的面向多样例粗粒度信息获取方法,FCM可以建立多细粒度概念与粗粒度概念之间的模糊认知关系,信息融合则用于构建粗粒度级概念的信息表达,NHL(Nonlinear Hebbian Learning)实现了基于数据源的自动学习,从而可以计算出粗粒度级概念的信息值,该方法在Fisher’s Iris公开数据集上分析并验证了有效性,并将此应用于基于科技文献大数据的科技人才评价发现中。

How to get coarse-grained information is one of the hot research topics in information management and infor-mation system. The paper presents an integrated algorithm of Fuzzy Cognitive Map（FCM）and information fusion of min-ing coarse-grained information for multi-instances. The fuzzy cognitive relations between multiple fine-grained concepts and a coarse-grained concept can be established by FCM. The information of coarse-grained concept can be constructed by information fusion. These fuzzy association values can be learned automatically directly from data resources by Non-linear Hebbian Learning（NHL）. The efficiency of method has been analyzed and demonstrated in the dataset of Fisher’s Iris and applied in scientific and technical talents evaluation based on the big data of scientific and technical literatures.