摘要
数据集市可为企业提供产品设计知识处理的廉价途径。因此,在建立数据集市的基础上,提出了一种产品设计知识处理方法,设计人员通过CBR系统输入查询请求,由OLAP通过钻取、切片等操作完成相似实例的匹配,通过知识约简技术对OLAP输出的相似实例及其属性进行简化。利用了粗集理论对属性进行重要性的判断和约简,为处理定量问题,给出了一种新的离散化方法,并利用相似学理论度量实例的相似性,以提高检索准确性和检索效率。最后,给出了一个的应用实例。
Data mart is a cheap method to give management analysis for product design knowledge processing. Building data mart of product design case for corporations, a knowledge processing method for product design was presented. Designers could input inquiries into Case-Based Reasoning （CBR） system. Then On-Line Analysis and Processing （OLAP） drilled down and found the similar case. Knowledge reduction techniques were adopted to reduce the retrieved similar cases output from OLAP, which improved CBR. Rough set theory was applied to calculate the important degree of each feature attribute and remove the redundant ones. And to deal with the quantitative features, a new discrete method was put forward, and the veracity and efficiency of case retrieval has been improved. The last an example was introduced.
出处
《制造业自动化》
北大核心
2008年第12期6-9,61,共5页
Manufacturing Automation
基金
南京人口管理干部学院重点科研项目基金(2007B04)
关键词
数据集市
知识管理
基于实例的推理
知识约简
data mart
knowledge processing
case-based reasoning
knowledge reduction
