一种融合专家和数据的综合知识建模框架.pdfVIP

浙江 工业大学学报 Vol. 35 No. 4 第 35 .4.第 4 期 2007 年8 月 JOURNAL OF ZHEJIANG UNlVERSITY OF TECHNOLQGY Aug. 2007 一种融合专家和数据的综合知识建模框架 张美玉，侯向军 {浙江工业大学 软件学院，浙江杭州 310032) 摘要: 专家系统过分依赖于专京的知识，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走模领拔的一个难题. 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 I F-THEN 旅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，通过来类分析、规则提取、近似榕理、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. 关键词:知识建模;专家知识 ;lF-THEN 规则;聚类分析;规则提取 中图分类号:TP182 文献标识码:A 文章编号:1006-4303(2007)04-0441-03 A knowledge-based modeling framework fusing expert knowledge and domain data ZH ANG Mei-yu , HOU Xiang-hui (College 01 Soltwar. , Zh.ji.ng University o[ T町阳。logy， Hangzhou 310032 , China) Abstract : Expert systems rely cxcessively o n expert knowledge. H ow to accurately obtain expcrt knowledge is a bottlen巳ck for the expert system and how to combine historical data and expert knowledge effectively is a difficult pr oblem in intelligent modeling domain. A fusion of expert knowledge and histo ri cal da ta integrated knowledge modeling framework is proposed . Firstly , a simple IF- THEN rule is u sed to r epresent the expert know lecige. Then a series of data analysis approaches are developed based on h istorical data , namely as data analysis engine in cluding clus- ter an alysis . rule extract ion , approximatc reasoning and rule adj ustment. This analysis engine can help experts to establish a nd optimiz.e the knowledge mode l correctly. l t can provide an inte- grated m od el ing method combining exper