基于CWM的元数据集成中形式化推理技术的研究  

Study on Formal Reasoning in Metadata Integration Based on CWM

作　　者：赵晓非[1] 黄志球[1] 

机构地区：[1]南京航空航天大学计算机科学与工程系,南京210016

出　　处：《计算机科学》2006年第12期177-182,共6页Computer Science

基　　金：航空科学基金(01I52058)资助。

摘　　要：在使用公共仓库元模型(CWM)进行元数据集成的过程中,参与集成的团体的不同经验以及描述数据的不同视角不可避免地带来元数据的冲突和冗余等问题,然而CWM的图形化特点使之缺乏精确的语义,所以如何在其上进行推理以自动发现这些问题至今没有得到很好的解决。本文研究了利用描述逻辑,一个一阶谓词逻辑的可判定子集,形式化CWM元模型和模型并进行推理的方法,将一致性检测分为水平一致性和演化一致性进行分别处理,在处理演化一致性的过程中对CWM元模型进行了扩展,使之支持元数据的版本能力从而能够推理由于演化引起的不一致问题,然后利用推理引擎LOOM对两种情形进行推理检测以发现不一致信息,取得了令人满意的结果,表明本文提出的方法是可行的。During the metadata integration based on Common Warehouse Metamodel（CWM）, the different experiences and views of describing data of organizations involved in integration bring metadata on some problems inevitably, such as ineonsistencies and redundancies. However, reasoning on CWM metadata for detecting these problems automatically is diffieuh because CWM metamodel and metadata lack precise semantics. In this paper, an approach is proposed to formalize and reason on CWM metamodel and metadata in terms of a logic belonging to Description Logics, which are subsets of First-Order Logic. The approach distinguishes consistency into horizontal consistency and evolution consistency. To evolution consistency, extends CWM metamodel with version capabilities so that reasoning about inconsistency caused by evolution can be done. Then reasoning engine LOOM is applied to check eonsisteney for above two situations, the results are encouraging.

关 键 词：公共仓库元模型(CWM) 描述逻辑 元数据集成 元数据演化 水平一致性 演化一致性 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]