A Survey on Ontology Mapping
[2006][SIGMOD][Namyoun Choi]
摘要:
Ontology日益被看作使得异构系统和SW应用之间交互的关键因素。
本文介绍了本体映射的分类,描述了每种分类的特点。比较了这些特点和测量工具、系统和每类映射的相关工作。
一、Introduction:
1、本体的独立开发这一特性导致对相同领域或者重叠领域的本体的定义有所不同,而这些独立的 本体开发团体之间缺乏充分沟通,所以:
——需要使用本体映射来促进他们之间的互操作。
2、本文将本体映射分为三大类[相关论文]:
1)综合的、全局本体 <----> 局部本体
——描述全局本体与局部本体之间的关系
[3]Learning to Match the Schemas of Data Sources: A Multistrategy Approach(2003)
[4]Synthesizing an Integrated Ontology(2003)
[1]Ontologies for Enterprise Knowledge Management(2003)
[7]A Framework for Ontology Integration(2001)
2)局部本体 <----> 局部本体
——使得在高度动态和分布的环境中的互操作成为可能
[6]C-OWL: Contextualizing Ontologies(2003)
[1]Ontologies for Enterprise Knowledge Management(2003)
[8]Semantic Coordination: A New Approach and an Application(2003)
[9]Learning to Map between Ontologies on the Semantic Web(2003)
[12]MAFRA – An Ontology Mapping FRAmework for the Semantic Web(2003)
[13]Resolving Terminological Heterogeneity in Ontologies(2002)
[14]Representing and reasoning about mappings between domain models(AAAI 2002)
3)mapping for 本体合并与联合
——被作为本体重用处理的一种途径
[15]PROMPT: Algorithm and Tool for Automated Ontology Merging and Alignment(AAAI2000)
[16]Ontomorph: A Translation System for Symbolic Knowledge(2000)
[17]FCA-Merge:Bottom-Up Merging of Ontologies(2001)
[18]Smart:Automated Support for Oontology Merging and Alignment(1999)
[19]Rule Induction for Concept Hierarchy Alignment((IJCAI 2001)
[20]Anchor-PROMPT: Using Non-Local Context for Semantic Matching((IJCAI 2001)
3、本文的内容:
基于详细的评价标准对这三类本体映射的工具和系统进行比较。这些标准是:
a、input requirements —— 需要的输入
b、level of user interaction —— 使用者交互的层次
c、type of output —— 输出的类型
d、content of output —— 输出的内容
e、the following five dimensions
structural, lexical, domain, instance-based knowledge, and type of result
——结构的、词汇的、领域的、基于场合的知识、结果类型
4、全文分为四大部分:
Section2:meanings of ontology mapping, ontology integration,merging, and alignment
Section3:
·characteristics and application domains of three different categories of ontology mapping are discussed.
·The tools, systems, frameworks, and related work of ontology mapping are surveyed based on the three different ontology mapping categories.
·Then the overall comparison of tools or systems about ontology mapping is presented
Section4:conclusion and presentation of future work
二、术语:本体映射,本体综合,合并,联合
1、本体映射在很多领域中被用到:ontology integration, merging, and alignment
——这三个领域中所用的的工具都被归为本体映射使用的工具。
还有一个与之相关的领域:schema matching,是数据库中的主要研究内容,本文何种不涉及
[3]Learning to Match the Schemas of Data Sources: A Multistrategy Approach(2003)
[36]A comparative analysis of methodologies for database schema integration(1986)
[37]Semantic Integration Research in the Database Community: A Brief Survey(2005)
[38]Corpus based schema matching(2005)
2、ontology integration, merging, and alignment——都可以被看成本体重用的一种方式
·ontology merging:
——从现存、不同的、并和同一主题相关的两个或者多个本体中生成单个、一致的本体的过程;
——其中被合并的那些本体是有相似性或者部分重叠的
·ontology integration:
——在两个或者多个现存的、不同的、并且描述不同主题的本体基础上生成单独的、在一个领域 中的本体
——原来的多个本体应该是相关的,在他们在被合并后的结果中,原有的本体中的内容会有变化
·ontology alignment:
——建立两个原始本体之间的链接
——本体联合的源本体是互相一致的、分离的;当需要充足的领域支持时才这么做。
3、ontology mapping(三种情况分别定义)
1)综合的、全局本体 <----> 局部本体
把存在于某一个本体中的一个概念映射到一个视图或者一个在其他本体之上的查询
2)局部本体 <----> 局部本体
在语义关系的基础上,把一个本体中的实体翻译为目标本体中的对应实体的过程。其中的源实体 和目标实体在语义概念上相关。
3)mapping for 本体合并与联合
在这种情况下,本体映射在源本体之间建立通信,并决定他们之间重叠的概念、同义词或者对其 他源本体来说独特的概念的集合。这种映射鉴别多个源本体之间的相似和不同之处,用于进行合 并或者联合。
三、本体映射的种类:
1、综合的、全局本体 <----> 局部本体
1.1 实力和缺点
优点:容易定义映射和找到映射的规则(因为有一个全局本体,全局本体提供共享的词汇)
缺点:缺乏可维护性和可测量性,因为局部本体的修改很容易影响到对全局本体的映射
该映射不能在包含不一致信息的相同或相似领域的不同本体之间进行,因为全局本体无法被创建
1.2 应用领域
SW、企业知识管理、信息/数据整合
1.3 工具、系统和相关工作
·LSD(learning source description):使用多策略学习方法的半自动创建语义映射。
包含多个Learner:分为两类 base learner 和 meta learner
——base learner:The Name Learner、Content Learner、Na?ve Bayes Learner、XML Learner
——meta learner:meta learner
映射过程分为两个阶段:
——training:a small set of data sources has been manually mapped to the mediated schema and is utilized to train the base learners and the meta learner
——matching:the trained learners predict mappings for new sources and match the schema of the new input source to the mediated schema
·MOMIS (Mediator Environment for Multiple Information Sources):
MOMIS creates a global virtual view (GVV) of information sources, independent of
their location or their data’s heterogeneity.
分为五个阶段:略
·A Framework for OIS(Ontology integration system):
——用描述逻辑来表达本体之间的映射query和本体
——使用两种方法:从全局本体的每一个概念 --->局部本体的概念(global-centric approach)
从局部本体的每一个概念 --->Global本体的概念(localcentric approach)
2、局部本体 <----> 局部本体 (是一个重点)
更适合Web上的高度动态的,开放的,分布式环境下的互操作。
2.1 实力和缺点
优点:适用与各局部本体因为包含不一致的信息而不能被综合或合并时,提供他们之间的互操作。
缺点:局部本体之间缺少共同的词汇.
2.2 应用领域
Web,Semantic web
2.3 工具、系统和相关工作
·Context OWL:Contextualizing Ontologies
·CTXMATCH:CTXMATCH is an algorithm for discovering semantic mappings across hierarchical classifications (HCs) using logical deduction.
·GLUE:semi-automatically creates ontology mapping using machine learning techniques
·MAFRA:Ontology MAapping FRAmework for distributed ontologies in the Semantic Web
provides a distributed mapping process that consists of five horizontal and four vertical modules.
·LOM:Lexicon-based Ontology Mapping
·QOM:Quick Ontology Mapping,a efficient method for identifying mappings between
two ontologies because it has lower run-time complexity.
·ONION:ONtology compositION system
·OKMS:Ontology-based knowledge management system。mapping is used for combining distributed and heterogeneous ontologies
·OMEN:Ontology Mapping Enhancer,OMEN is a probabilistic ontology mapping tool which enhances the quality of existing ontology mappings using a Bayesian Net
·P2P ontology mapping:This work proposes the framework which allows agents to interact with other agents efficiently based on the dynamic mapping of only the portion of ontologies relevant to the interaction.
3、mapping for 本体合并与联合
可以通过合并处理创建一个一致的本体。It also creates links between local ontologies while they remain separate during the ontology alignment process.
映射不存在于合并后得到的的本体 与 被合并的若干局部本体之间,
但存在于被合并的若干局部本体之间。
第一步:找出待合并或者待联合的多个局部本体之间的相似和冲突。
3.1 实力和缺点
3.2 应用领域
Many applications such as standard search, e-commerce, government intelligence, medicine, etc., have large-scale ontologies and require the reuse of ontology merging processes。
3.3 工具、系统和相关工作
·SMAR:SMART is a semi-automatic ontology merging and alignment tool.
·PROMPT:PROMPT is a semi-automatic ontology merging and alignment tool.
·OntoMorph: OntoMorph provides a powerful rule language for specifying mappings, and facilitates ontology merging and the rapid generation of knowledge-base translators.
·HICAL (Hierarchical Concept Alignment system):
provides concept hierarchy management for ontology merging/alignment (one concept hierarchy is aligned with another concept in another concept hierarchy), uses a machine-learning method for aligning multiple concept hierarchies, and exploits the data instances in the overlap between the two taxonomies to infer mappings.
·Anchor-PROMPT:takes a set of anchors (pairs of related terms) from the source
ontologies and traverses the paths between the anchors in the source ontologies.
·CMS (CROSI Mapping System): CMS is an ontology alignment system.
·FCA-Merge:a method for ontology merging based on Ganter and Wille’s formal
concept analysis, lattice exploration, and instances of
ontologies to be merged.
·CHIMAERA: CHIMAERA is an interactive ontology merging tool based on the Ontolingual
ontology editor.
4、本体映射工具和系统的比较
四、结论
[文中的一个图]
下面要重点看的五篇论文:
A、[2002][www][ AH Doan][Learning to Map Between ontologies on the semantic web]
B、[2002][EKAW][Alexander Maedche][MAFRA— A MApping FRAmework for Distributed Ontologies]
C、[2003][IEEE Internet Computing][D Beneventano][Synthesizing an integrated ontology](未打)
D、[2004][ISWC][John Li][QOM - Quick Ontology Mapping](未打)
E、[2003][IEEE Intelligent Systems][A Maedche][Ontologies for enterprise knowledge management](未打)
联系客服