An Ontology Alignment Validation Approach Based on Supervised Machine Learning Algorithms and Automatic Schema Matching Approach
Faten Abbassi, Yousra Bendaly Hlaoui · 2024
The existence of various representations of the same ontology poses a challenge in terms of manipulating knowledge across different computational domains. To address this issue, it would be prudent to harmonise similar ontologies by reducing their level of heterogeneity. This proposed solution involves aligning comparable ontologies through the utilisation of established ontology schema-matching techniques. In this paper, we introduce an approach for ontology alignment that leverages these techniques along with machine learning algorithms. To accomplish this, we propose a method for constructing a matrix based on ontology matching techniques, specifically employing element matching and structure matching techniques facilitated by elementary matchers. Subsequently, once the matrix is constructed, we use a composite matcher as a classifier to determine the degree of similarity between the two ontologies. Since these matchers are not readily available, we put forth the idea of implementing them in this paper using various supervised machine learning algorithms such as Neural Network and Logistic Regression. To validate our approach, we conducted an empirical assessment using the Conference track and the benchmark track of the reference ontologies provided by the Ontology Alignment Evaluation Initiative (OAEI 1).