大型知识库知识存储结构的研究及其在知识发现系统中的应用

在基于数据库和知识库的知识发现系统(KDD&K)的研究中,需对知识库中的重复、冗余、矛盾、循环的知识进行实时校验、修改,并能够发现知识短缺,指导KDD过程进行聚焦;在KDK过程中,需要找出有关联的知识组成的知识域以便于归纳、解释等具体应用需求,针对于此,该文提出了一种基于知识节点(属性)的图矩阵、二维链表、产生式规则的三级管理模式和数据存储结构,通过知识库管理系统(KBMS)实现了二层逻辑结构和一层物理结构的三层独立映射关系,大大压缩了知识的搜索空间。经在KDD&K原型系统中的具体应用,该知识库系统结构的定义以及相应的KBMS完全满足上述要求,并可推广至通用的大、中型知识库系统。

Study on Storage Structure of Knowledge Base and its Application in Knowledge Discovery System

On the study of knowledge discovery in database and knowledge base(KDD&K)system,it is required to verify and modify the knowledge in knowledge base(KB)on real time because of repetition,redundancy,paradoxicalness and circulation.It is also required to discovery the scarcity of knowledge to instruct the KDD task.During the course of KDK,it need find the co-relationship of knowledge nods to meet the demand of application of induction,explanation and so on.In order to solve these problems ,this paper presents a knowledge base management model and a data storage structure that consist of graph matrix,two-dimension chain and productive rules based on knowledge nodes.By these the searching space of knowledge is reduced sharply.The paper realize s this structure in a KDD&K prototype system.The result shows this management model and relevant KBMS can perfectly meet the demands said above.And it can be extended to universal large or middle scale knowledge base system.