基于描述逻辑及符号算法的事例相似性研究

项目名称： 基于描述逻辑及符号算法的事例相似性研究

项目编号： No.61262030

项目类型： 地区科学基金项目

立项/批准年度： 2013

项目学科： 自动化技术、计算机技术

项目作者： 古天龙

作者单位： 桂林电子科技大学

项目金额： 47万元

中文摘要： 从知识密集型事例推理（CBR）应用的需求出发，研究描述逻辑、时序描述逻辑和模糊描述逻辑的事例表示和推理机制，建立适合于知识密集型CBR应用中事例表示的描述逻辑C-fTDL(x)；以有序二叉决策图（OBDD）、代数决策图（ADD）及其扩展形式为基础，研究C-fTDL(x)中TBox、RBox和ABox符号表示的新型xDD结构，建立C-fTDL(x)的符号xDD推理方法及算法；以描述逻辑ALC、ALCNR和SHI的事例相似性为基础，探讨模糊和时序信息综合的相似性机理，建立能够度量C-fTDL(x)语义和语法结构的事例相似性度量方法；以图/子图结构匹配的符号OBDD/ADD算法为基础，研究C-fTDL(x)下事例检索的符号xDD算法；以产品装配规划、软件复用为背景，研究知识密集型CBR应用中本体的构建和集成机制，开发领域无关本体原型系统。以期为知识密集型CBR中事例相似性研究提供新的理论和方法。

中文关键词： 事例推理；描述逻辑；有序二叉决策图；相似性度量；符号算法

英文摘要： In terms of the knowledge intensive case-based reasoning (CBR) applications, the case representation and reasoning mechanism of description logic, temporal description logic and fuzzy description logic will be studied so that the description logic family DL(x) will be proposed in order to represent the knowledge in the knowledge intensive CBR applications. In the framework of ordered binary decision diagram (OBDD), algebraic decision diagram (ADD) and their variants, the novel structure xDD to symbolically represent TBox, RBox and ABox of C-fTDL(x) will be explored so as to develop the symbolic reasoning methods and algorithms for description logic C-fTDL(x). On the basis of case similarity assessment using description logic ALC, ALCNR and SHI, the integrated similarity assessment mechanism of cases containing fuzzy and temporal information will be probed, and the case similarity assessment methodology for description logic C-fTDL(x) will be proposed, where the rich semantics and syntactic knowledge could be involved. Through analyzing the OBDD/ADD based symbolic algorithms for the matching between two graphs or sub-graphs, the similarity-based retrieval algorithms in description logic C-fTDL(x) will be developed. From the applicable backgrounds of product assemble planning and software reuse, the domain indep

英文关键词： case-based reasoning；description logic；ordered binary decision diagram；similarity assessment；symbolic algorithm