Article Preview
TopIntroduction
Ontologies play an important role in many computer applications where it is necessary to overcome the problems of heterogeneity and diversity in semantics. They define a formal semantics for the information enabling the interoperability of information sources (Fensel, 2001). With the emergence of the semantic web, ontologies have proliferated and are accessible to a wide audience. This leads to the appearance of several, but overlapping ontologies for the same domain and each source of information is free to choose the most appropriate ontology to its needs. Consequently, the problem of information interoperability turns to ontology interoperability. Ontology interoperability can be achieved by ontology matching tools which aim at finding semantic correspondences between related entities of different ontologies (Euzenat & Shvaiko, 2013). These correspondences express semantic relations between entities of different ontologies. The set of these correspondences constitutes an alignment between ontologies. An alignment is used to import data from an ontology to another, translating queries between them or merging ontologies in a global one (Euzenat & Shvaiko, 2013).
Ontologies are not static conceptual models of “eternal” truth, but artifacts reflecting our gradual understanding of reality (Hepp, 2007). Hence, ontologies are continuously in evolution so that they reflect changes in the application domain or in the business strategy and incorporating additional functionality according to changes in users’ needs (Stojanovic, 2004). The dynamic aspect of ontologies can affect and make obsolete the alignment between them. This is a case of the known alignment evolution problem (Euzenat & Shvaiko, 2013; Dos Reis et al., 2013; Groß et al., 2013; Dos Reis et al., 2015; Martins & Silva, 2009). We call such a case as alignment evolution under ontology change.
Usually, alignment evolution corresponds to the creation of a new alignment, derived from an existing one (Euzenat & Shvaiko, 2013). Recently, some approaches (Dos Reis et al., 2013; Groß et al., 2013; Martins and Silva, 2009) have emerged to deal with alignment evolution under ontology change. The main challenge of these approaches is how to adapt the old alignment according to ontology change. Influenced by the underlying representation of ontologies, none of these approaches considers the formal semantics of ontologies and hence of alignment. Therefore, many properties about alignment quality are neglected when dealing with this problem. In practice, knowledge systems encode ontologies and alignments as knowledge bases (Grimm & al., 2011). The set of axioms contained in these bases constitutes the explicit knowledge and implicit knowledges are logical consequences of them. Under this logical representation of ontologies and alignment, ontology evolution may result in serious consequences for the alignment. For instance, some added axioms may lead to inconsistent alignment. Moreover, removed axioms may still be logical consequences of alignment while breaking users’ needs and applications’ requirements.