节点文献

不完备模糊目标信息系统的规则提取研究

Research on Rule Rule Acquisition in Incomplete and Fuzzy Objective Information Systems

【作者】 梁小芝

【导师】 王伟平;

【作者基本信息】 中南大学 , 计算机应用技术, 2008, 硕士

【摘要】 经典粗糙集理论是完备信息系统的数据分析与处理的重要工具之一,其主要思想是在保持分类能力不变的前提下,通过知识约简,进而导出问题的分类规则或决策规则。但是完备信息系统只是一种理想化的情况,而在实际问题中,往往会遇到条件属性值域不完备,目标属性值域是模糊的情形,,本文称之为不完备模糊目标信息系统(Incomplete and Fuzzy Objective Information System,简称IFOIS),对它的研究具有重要的现实意义。本文以IFOIS作为研究对象,以粗糙集理论和模糊集理论为工具,研究不完备模糊目标信息系统的知识获取方法:以基于改进型相容关系、限制容差关系、对称相似关系下不完备信息系统的粗糙集模型为基础,进一步拓展到不完备模糊目标信息系统,建立了基于改进型相容关系、限制容差关系、对称相似关系下的不完备模糊目标信息系统的粗糙集扩展模型;基于以精度相等思想,讨论了各种模型下不完备模糊目标信息系统知识约简方法,给出了基于改进型相容关系的(α,β)精度约简算法、基于限制容差关系的β-上协调约简算法(α-下协调约简算法),基于对称相似关系的(α,β)协调约简分辨矩阵算法;最后研究了基于相容关系、改进型相容关系、限制容差关系、对称相似关系等四种二元关系下不完备模糊决策信息系统的决策规则提取方法,给出了相应的规则提取算法,并通过实例验证了方法是可行的。

【Abstract】 Classical Pawlak rough set theory is one of important tools for data analyzing and processing in complete information system. The main idea is to deduce the classification rules or decision-making rules by knowledge reduction on the conditions of the same capability of classification. In fact, the complete information system is only an ideal situation. Because conditional attributes may be incomplete, objective attributes may appear ambiguous situations, which is called the rough set model of incomplete and fuzzy objectives information system (It is called IFOIS for short) in this paper. So, it has very important practical significance to research IFOIS.In this paper, addressing IFOIS, we research the knowledge acquisition approach based on rough set theory and fuzzy set theory. The expanded rough set models of IFOIS are established under the condition of improved-tolerance relation, limited-tolerance relation and symmetric similarity relation. Furthermore, we discussed all kinds of knowledge reduction approaches among different models and proposed three advanced algorithms, which include precision reduction algorithm which is (α,β) binary relation under improve-tolerance relation,β-upper precision reduced (orα-lower precision reduced) under limited-tolerance relation and (α,β) coordinated reduction resolution matrix algorithm under symmetric similarity relation. At last, through analysising extraction of the decision-making rules of IFOIS for four different binary relations, we proposed the corresponding rule extraction algorithms and proved their feasibility by some examples .

  • 【网络出版投稿人】 中南大学
  • 【网络出版年期】2009年 01期
  • 【分类号】TP18
  • 【被引频次】2
  • 【下载频次】115
节点文献中: 

本文链接的文献网络图示:

本文的引文网络