一种融合专家和数据的综合知识建模框架  

A knowledge-based modeling framework fusing expert knowledge and domain data

作　　者：张美玉[1] 侯向辉[1] 

机构地区：[1]浙江工业大学软件学院,浙江杭州310032

出　　处：《浙江工业大学学报》2007年第4期441-443,共3页Journal of Zhejiang University of Technology

基　　金：国家自然科学基金资助项目(60603087);浙江省科技计划项目(2006C33045)

摘　　要：专家系统过分依赖于专家的知识,如何准确地获取专家知识成为系统工作的瓶颈,将历史数据和专家知识有效结合成为智能建模领域的一个难题.提出一个融合专家知识和历史数据的综合知识建模框架,首先利用简单的IF-THEN推理规则表示专家的经验知识,再基于历史数据设计一系列自动分析方法,通过聚类分析、规则提取、近似推理、规则调整等过程形成一个数据分析引擎,协助专家正确建立并优化知识模型,为领域专家提供一个将个人知识和数据抽取规则相融合的综合建模方法.Expert systems rely excessively on expert knowledge. How to accurately obtain expert knowledge is a bottleneck for the expert system and how to combine historical data and expert knowledge effectively is a difficult problem in intelligent modeling domain. A fusion of expert knowledge and historical data integrated knowledge modeling framework is proposed. Firstly, a simple IF-THEN rule is used to represent the expert knowledge. Then a series of data analysis approaches are developed based on historical data, namely as data analysis engine including cluster analysis, rule extraction, approximate reasoning and rule adjustment. This analysis engine can help experts to establish and optimize the knowledge model correctly. It can provide an integrated modeling method combining expert knowledge and data extraction rules for domain experts.

关 键 词：知识建模 专家知识 IF-THEN规则 聚类分析 规则提取 

分 类 号：TP182[自动化与计算机技术—控制理论与控制工程]