Abstract
Different repositories of ontology are available on the web to share common understandings of the knowledge of different domains with semantic web applications. They store, index, organize, and share ontologies and alignments between them that allow applications to search for and use the appropriate semantics on the fly. The quality of the ontologies and the alignments between them is a great challenge to guarantee the usefulness of ontology repositories. Like ontologies, alignments are subject to changes throughout their life cycle, which can decrease their quality. As a result, alignments must be evolved and maintained in order to keep up with the change in ontology or to meet the demands of applications and users. This chapter reviews and classifies the main ontology alignment change approaches. In addition, the chapter presents a new approach for the alignment change problem. The approach proposes a general framework that consists of a process of change. Various methods, each with a specific purpose, are proposed to automate and support the change process.
TopIntroduction
Ontologies and alignments between them enhance semantic interoperability for many Web applications such as Web services interoperability and query answering on the Web to name a few (Euzenat & Shvaiko, 2013). While ontologies provide specifications of the semantics of vocabularies in order to share a common understanding of domain knowledges, alignments overcome heterogeneity and diversity in these specifications.
With the emergence of the semantic Web, ontology repositories such as Swoogle (Ding et al, 2004), Watson (D'Aquin,, Gridinoc,, Angeletou,, Sabou,, & Motta, 2007), OntoSelect (Buitelaar, Eigner,, & Declerck, 2004), the DAML ontology library (http://alignapi.gforge.inria.fr/aserv.html) consider alignments as first class objects, enhancing the dynamic interoperability of ontologies. They, store, index, organize and share alignments. These infrastructures allow applications to seek and use on the fly the appropriate alignments.
The quality of ontologies and alignments between them is a great challenge to guarantee the usefulness of these repositories. Like ontologies, alignments are subject to changes throughout their life cycle, which can decrease their quality. Many reasons can trigger this change. Alignments cannot keep their consistency in time because of the dynamicity of ontologies. For instance, adding new knowledge in ontologies can make alignments inconsistent (Euzenat, 2015). Retracting knowledge from ontologies in response to some needs forces also alignments to follow this change. Another reason that can trigger the alignment change is alignment debugging and repair. Indeed, ontology matching tools may produce redundant, missed, or erroneous correspondences that can lead to an alignment inconsistency (Wang & Xu, 2008; Jean-Mary, Shironoshita, & Kabuka, 2009; Meilicke & Stuckenschmisdt, 2009; Meilicke & Stuckenschmisdt, 2007; Qi, Ji, & Haase, 2009).
Key Terms in this Chapter
Belief Base Revision: Is the process of changing a finite set of beliefs to take into account a new belief.
Alignment Evolution: Is the task of alignment change after delivery in order to keep up with the change in ontology or to meet the demands of applications and users.
Ontology Alignment: A set of correspondences stating how entities of different ontologies semantically are related.
Alignment Debugging: Is the task of alignment change before alignment delivery to diagnose and repair ontology alignments produced by matching tools.
Alignment Revision: Is the task of change of an alignment between ontologies as logical theories.
Alignment Change: Any modification on the alignment correspondences.