利用本体云影模型的混合本体方法

Ontology cloud-shadow model based on hybrid ontology approach

近几年,使用本体作为语义整合和互操作性的解决方案,在规模巨大、动态、异构的环境,混合本体方法相比较单本体方法、多本体方法有着一定的优势.本体云影模型能更好的体现知识的特征,通过将本体云模型跟隶属云相结合,将隶属云作为概念的原子模型,本体云作为知识结构的原子模型,两者结合提供了从概念到领域知识一致的表达,云影模型跟语义、语用紧密相关,能表达知识的不确定性、不协调性、时变性及一定的群体性、规律性,体现了语用和新知识观,从而能孕育生命力和创造性.本文基于本体云影模型提出了一种混合本体方法.基于本体云影模型构建的混合本体方法结构上可以看成是聚类方法跟"联岛"方法的一个结合,内容上也结合了两者的优点.能更好的体现知识的特征——模糊性、统计性、不确定性;反应知识的粒度;能解决其他方法的一些问题,可以很好的应用于语义整合领域.

In a highly dynamic heterogeneous environment,uncertainty,abnormality and inconsistency are becoming common properties of ontological knowledge.For many web applications,dealing with vague,incomplete and even inconsistent knowledge is very hard.In new knowledge view,uncertainty and inconsistency is not an obstacle to using knowledge but leading to creativity of world.Based on this view,an ontology cloud shadow model（OCSM） is proposed,which combines linguistic cloud model（LCM） with ontology cloud model（OCM）.The linguistic cloud acts as the atomic unit of concept and the ontology cloud acts as that of knowledge.Thus,knowledge construction interacts with concept construction in a coherent manner.Both models embody the uncertainty,abnormality and inconsistency very well,and,as can be imagined,cloud and its shadow is vague and time-varying,which is very similar to the properties of ontological knowledge.Ontology-based semantic interoperability and integration approaches can be classified into： the single ontology approach,the multi-ontology approach,and the hybrid approach.In the first approach,all data sources are related to global ontology.In the second approach,each data source has its own ontology.The third approach is a combination of the two previous ones.In this paper,a hybrid ontology mediation approach based on the OCSM is presented.This approach has several steps： ontologies in same fields are first integrated into a large ontology cloud;ontology clouds in different fields are then merged together to create an OCSM;finally,the model is used to solve semantic divergence.Our approach can be regarded as a combination of the clustering-based hybrid approach and the island-based one.It is more flexible than the single ontology approach and the multi-ontology approach.Comparing with other hybrid approaches,it has a flexible hierarchical level to reflect the knowledge granularity,and sufficient definitions for complex operations,such as ontology mapping,integration,and merging.Furthermore,knowledge modeling,managing,data-mining and evolving can be also tackled in this approach.It has been tested in an ontology engineering project,and the results show significant improvements over other techniques.