基于规则面向属性的数据库归纳的无回溯算法  

A Nonbacktracking Algorithm for the Rule Based Attribute oriented Database Induction Approach

作　　者：周生炳[1] 张钹[2] 成栋[3] 

机构地区：[1]空军电讯工程学院,西安710077 [2]清华大学计算机科学与技术系,北京100084 [3]中国人民大学工商管理学院,北京100872

出　　处：《软件学报》1999年第7期673-678,共6页Journal of Software

基　　金：国家自然科学基金;中国博士后科学基金

摘　　要：该文提出了基于规则的面向属性知识发现方法的无回溯算法．把背景知识理解为特殊的逻辑程序，并把它的子句展开为完全归结子句，然后按照用户要求，定义并确定每个属性的恰当层次．每个属性的多个值归纳为恰当层次中的值，只需一遍扫描。In this paper, a nonbacktracking algorithm is presented for the rule based attribute oriented database induction approach. The background knowledge is considered as a limited logic program, and every clause of it is expanded as a complete resolution clause. Then, the concept of exact level for an attribute in the concept hierarchies is defined and specified according to its threshold required by the users. The values of each attribute are inducted as ones in its exact level by only one pass without backtracking.

关 键 词：知识发现 数据库 归纳 无回溯算法 数据采掘 

分 类 号：TP311.13[自动化与计算机技术—计算机软件与理论]