论文部分内容阅读
不完全信息下的近似推理是知识工程面临的困难问题之一.文章提出了一种具有非单调性质的优先逻辑程序理论.该理论能够对知识的解释进行综合评判,进而优选解释,使其成为现有知识的最佳理论逼近,达到在择优意义下的理论完全化,避免了对知识的完全性及一致性要求.为获取应用领域的优先逻辑程序,基于归纳逻辑程序设计技术设计了一种多方法归纳学习算法,该算法具有较强的归纳能力.此理论与算法已应用在863农业专家系统中,并获得满意结果
Approximate reasoning under incomplete information is one of the difficult problems facing knowledge engineering. This paper presents a theory of non-monotonic priority logic program. This theory can comprehensively judge the explanation of knowledge and then explain it optimally, making it the best theoretical approximation of existing knowledge, achieving the completeness of theory under the preferential meaning and avoiding the requirement of completeness and consistency of knowledge. In order to obtain the priority logic program in application domain, a multi-method inductive learning algorithm is designed based on inductive logic programming technology, which has strong inductive ability. This theory and algorithm has been applied in 863 agricultural expert systems and obtained satisfactory results