多概念层次上基于赋范划分距离的分类算法  

Classify Algorithm Based on Normal Partition Distance at Multiple Concept Levels

作　　者：许骏[1] 柳泉波[1] 

机构地区：[1]北京师范大学现代教育技术研究所,北京100875

出　　处：《计算机工程》2001年第7期67-68,117,共3页Computer Engineering

摘　　要：应用知识发现理论与方法从大型考生信息数据库中提取隐含的、前所未知的、对决策有潜在价值的知识和规则,是当前计算机教育 应用的新课题。根据网上远程考试与评价系统的特点,讨论了挖掘的目标规则类型及其挖掘算法。针对算法的不足,提出了一种新的分ID3类算法多概念层次上基于赋范划分距离的分类算法,简要介绍了在知识发现过程中的应用。Application of KDD technology to extract implicit,previously unknown,and potentially useful rules from very large databases withexaminees' information is a new task in the computer eductional application category.Based on the characteristics of distance test and evaluation system,we first discuss the result rules' kinds and relative mining algorithms.Then a new classify algorithm based on normal partition distance at multiple concept levels is addressed to overcome ID3's shortcoming.Finally application of XML in the complete process of KDD is briefly introduced.

关 键 词：知识发现 数据挖掘 赋范划分距离 分类算法 数据库 远程考试系统 INTERNET 

分 类 号：TP311.13[自动化与计算机技术—计算机软件与理论] G434[自动化与计算机技术—计算机科学与技术]