摘要
给出了依据决策树进行知识发现的问题求解过程、策略与技术关键,提出了利用加权熵来进行决策树结构优化的方法,结合结构选型的特征对决策树基本算法ID3进行了改进;以20个高层结构实例为基础,给出了高层结构智能选型分类规则挖掘的实例及其所发现的部分选型知识;获取了1008个高层结构工程实例,建立了基于实例库与改进决策树方法的知识发现系统,为从工程实例中挖掘结构选型知识提供了方法与手段支持.实践表明,决策树方法能有效地从工程实例中开采隐含的结构选型知识,它将对充分地利用丰富的实例资源,提高结构选型设计质量与效率具有重要意义,也为解决结构智能选型知识获取“瓶颈”问题提供了一个新途径.
First, the settlement course, tactics, and technological key for KDD based on decision tree are addressed. The method of optimize structure that from make use of weighting entropy is Put forward. And with characteristic of structure from selection, the basic algorithm method of ID3 is improved. Secondly, based on 20 High-rise structure cases, the cases of High-rise structures intelligent from selection assorted case-mining and part of selection intelligent from which are addressed. Finally, 1008 high-rise building cases are collected. The KDD system which based on case-base and improve the method of decision tree is set up. And it will help the methods and measures sustain for mining structure form-selection from building cases. The given examples proved that the decision tree methods can exploit cryptic structure form-selection knowledge from building cases effectively, it has significant meaning for utilizing abundant case resources fully, improving the design of quality and efficiency structure form-selection, and offering a new approach for solve the problem of 'bottle-neck' structure intelligent form selection.
出处
《哈尔滨工业大学学报》
EI
CAS
CSCD
北大核心
2005年第4期451-454,567,共5页
Journal of Harbin Institute of Technology
基金
国家自然科学基金"九五"重大资助项目(59895410)
关键词
高层结构
智能选型
知识发现
决策树
high-rise structures
intelligent form selection
knowledge discovery in databases
decision tree
