基于知识粒度的不完备决策表的属性约简的矩阵算法

通过考虑云计算环境下的服务器资源负载严重、资源描述方式单一以及资源配置方法效率不高的问题以及云计算本身的特点,根据本体论和多代理机制提出一种高效实用的基于域(DOMAIN)和虚拟组织(Virtual Organiza-tion i.e.VO)的完全分布的、高效的资源配置策略。在这个策略中,每个节点都采用CCSM(Cloud Computing SystemModel)体系结构,该体系结构根据本体论引入了多代理的分层结构,并利用本体论表达语义信息来对资源进行管理,根据分层结构可以减少远程访问的时间,提高访问效率。根据语义信息来获得对概念的统一理解形式,过滤掉无关信息以缩小查询范围,并利用概念关联进行推理以实现对查询请求的泛化,由此可以提高查询过程中的查全率和查准率。

Discernibility Matrix Algorithm of Attribute Reduction Based on Knowledge Granulaion

Considering the disadvantages of traditional Cloud Computing system and the characteristics of Cloud Computing,according to the Multi-Agent principle and ontology,a high-effect resource management strategy which adopts the completely distributed structure based on DOMAIN and virtual organization(VO) was proposed.In the strategy,each of metadata system adopts the architecture called CCSM(Cloud Computing System Model),which adopts Multi-Agent Hierarchical structure and semantic method to manage service resource.According to semantic information,CCSM can obtain the consistent concept of query,reduce the inquiry scope,and realize the extensive request of query based on concept connection,which can enhance precision and recall in inquiring process.