一种基于模板的子句学习算法

A Clause Learning Algorithm Based on Template

针对归纳逻辑程序设计中学习子句所遇到的较大搜索空间问题,提出子句模板的概念.用遗传算法先学习子句模板,再结合标记矩阵和按信息增益抽样的方法将其转化为子句.设计了相应的适应度函数及遗传算子.理论分析与实验对比表明,该算法可有效缩小搜索空间,提高学习效率,并且可以学习递归子句,是一种有效的子句学习算法.

For the larger search space when learning clause in Inductive Logic Programming,we defined the clause template. Firstly,we learn the clause templates by Genetic Algorithm,and then convert it to the requisite clauses by combining tag matrix and information gain sampling.We designed the corresponding fitness function and genetic operators.Theoretical analysis and experiment comparison show that this algorithm can reduce the search space,improve the search efficiency and can learn recursion clause.It is an effective clause learning algorithm.