基于数据资源的认知图挖掘系统研究  

Study on the mining system for cognitive maps based on data resources

作　　者：李嫄源[1] 张国[2] 

机构地区：[1]重庆邮电大学科学技术处,重庆400065 [2]重庆理工大学计算机科学与工程学院,重庆400054

出　　处：《重庆邮电大学学报（自然科学版）》2011年第3期374-378,共5页Journal of Chongqing University of Posts and Telecommunications(Natural Science Edition)

基　　金：重庆市自然科学基金(CSTC;2008BB2065)~~

摘　　要：认知图是一种知识表达和管理工具,被广泛应用于知识表达和决策制定。传统的认知图建立方法大多依赖于领域内的专家经验,而专家经验往往在一定程度上存在知识的局限性、观点的偏向性等不足,因而通过专家经验建立的认知图可能与现实世界存在着较大的差距。构建了一个基于数据资源的认知图挖掘系统,通过对客观数据资源进行训练来建立认知图,摆脱对专家经验的依赖。该系统主要由数据资源预处理、认知图挖掘、认知图推理等部分组成,在对客观样本数据进行学习并建立认知图的基础上,通过认知图推理机制进行决策分析,最终实现对现实世界的真实模拟。最后通过对客观数据样本进行试验,证明该系统所挖掘出的认知图以及推理分析结果与现实世界基本一致。Cognitive maps are effective knowledge management tools and have been widely used in knowledge representation and decision-making.The traditional methods to establish them,which largely rely on expert experiences,inevitably have certain defects caused by such problems as limited knowledge and partiality of the experts concerned.Consequently the cognitive maps made in this way may fail to represent the reality.In order to avoid these defects,this paper proposes a data-based mining system which establishes cognitive maps by training data samples.The system mainly consists of three steps： data preprocessing,mining for cognitive maps and reasoning.Namely,it starts with analysis of sample data,then establishes cognitive maps accordingly and makes decision with the help of cognitive maps reasoning.Experimental results prove that the cognitive maps mined by this system are essentially true to the reality.

关 键 词：认知图 数据资源 认知图推理 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]