Alignment Conservativity Under the Ontology Change

Alignment Conservativity Under the Ontology Change

Yahia Atig, Ahmed Zahaf, Djelloul Bouchiha, Mimoun Malki
Copyright: © 2022 |Pages: 19
DOI: 10.4018/JITR.299923
Article PDF Download
Open access articles are freely available for download

Abstract

Recently, many methods have appeared to solve the problem of the evolution of alignment under the change of ontologies. The main challenge for them is to maintain consistency of alignment after applying the change. An alignment is consistent if and only if the ontologies remain consistent even when used in conjunction with the alignment. The objective of this work is to take a step forward by considering the alignment evolution according to the conservativity principle under the change of ontologies. In this context, an alignment is conservative if the ontological change should not introduce new semantic relationships between concepts from one of the input ontologies. The authors give methods for the conservativity violation detection and repair under the change of ontologies and they carry out an experiment on a dataset adapted from the Ontology Alignment Evaluation Initiative. The experiment demonstrates both the practical applicability of the proposed approach and shows the limits of the alignment evolution methods compared to the alignment conservativity under the change of ontologies.
Article Preview
Top

Introduction

The proliferation in the development of ontologies has led to the appearance of ontology repositories such as Bioportal1 and AgroPortal2 to store and share ontologies and alignments. The utility of these repositories depends not only on ontologies quality, but also on alignments between them. Indeed, the evolution of ontologies following changes in knowledge domains may affect and make obsolete the alignment between them. Thus, alignments must be evolved and maintained in order to keep up with the change in ontologies.

Recently, many methods have appeared to solve the problem of alignment evolution under ontology change (Zahaf, 2017). The main challenge for them is to maintain the alignment consistency after applying the change. An alignment is consistent if and only if the ontologies remain consistent even when used in conjunction with the alignment. The notion of consistency is approached by alignment evolution methods according to two different levels: structural consistency and logical consistency. The structural consistency ensures that the alignment obeys the constraints of its underlying representation structure (Martins & Silva, 2009). The logical consistency considers the semantics of the alignment, meaning that it does not introduce contradicting knowledge in ontologies (Euzenat, 2015; Zahaf & Malki, 2016).

The objective of this work is to take a step forward by considering the alignment evolution according to the conservativity principle under ontological changes. Jiménez-Ruiz et al. (2011) proposed the conservativity as a principle to minimize the number of potentially unintended logical consequences in the alignment. They state that mappings must not introduce new semantic relationships between concepts of one of the input ontologies. In other words, alignment should allow interaction between ontologies rather than providing a new description of the domain. In the context of alignment evolution under ontology change, an alignment is conservative if the ontological change should not introduce new semantic relationships between concepts from one of the input ontologies. We consider these relationships as violations of the conservativity principle following ontological changes.

To our knowledge, the problem of Alignment Conservativity under Ontology Change has not been studied yet. We built on top of previous work on repairing automatically evolved alignments under ontology change (Zahaf & Malki, 2016) and extend this work in several directions:

Complete Article List

Search this Journal:
Reset
Volume 16: 1 Issue (2024): Forthcoming, Available for Pre-Order
Volume 15: 6 Issues (2022): 1 Released, 5 Forthcoming
Volume 14: 4 Issues (2021)
Volume 13: 4 Issues (2020)
Volume 12: 4 Issues (2019)
Volume 11: 4 Issues (2018)
Volume 10: 4 Issues (2017)
Volume 9: 4 Issues (2016)
Volume 8: 4 Issues (2015)
Volume 7: 4 Issues (2014)
Volume 6: 4 Issues (2013)
Volume 5: 4 Issues (2012)
Volume 4: 4 Issues (2011)
Volume 3: 4 Issues (2010)
Volume 2: 4 Issues (2009)
Volume 1: 4 Issues (2008)
View Complete Journal Contents Listing