Learning Description Logic Ontologies: Five Approaches. Where Do They Stand?

Abstract

The quest for acquiring a formal representation of the knowledge of a domain of interest has attracted researchers with various backgrounds into a diverse field called ontology learning. We highlight classical machine learning and data mining approaches that have been proposed for (semi-)automating the creation of description logic (DL) ontologies. These are based on association rule mining, formal concept analysis, inductive logic programming, computational learning theory, and neural networks. We provide an overview of each approach and how it has been adapted for dealing with DL ontologies. Finally, we discuss the benefits and limitations of each of them for learning DL ontologies.

Learning multiple concepts in description logic through three perspectives

“…Since concepts are the basic component of a terminology, the first step is to conceive how to automatically learn a concept from available data. A number of algorithms together with implementations have been proposed in the literature to learn concepts in DL (Iannone et al, 2007;Fanizzi et al, 2008;Lehmann, 2009;Lehmann & Hitzler, 2010;Lima et al, 2018;Ozaki, 2020). They are inspired on techniques developed within the field of Inductive Learning Programming (ILP) (De Raedt, 2008), whose general goal is to automatically induce logic programmes from data.…”

“…Thus, a number of approaches to learn ontologies written in DL have been proposed in the last years (Iannone et al, 2007;Fanizzi et al, 2008;Lehmann & Hitzler, 2010;Bühmann et al, 2016;Konev et al, 2017a;Funk et al, 2019;Ozaki, 2020). However, the majority of them follow a single concept learning strategy, that is, they focus on learning piecewise concepts.…”

Ontologies and Data Management: A Brief Survey

“…-Learning (ontologies and queries): see the survey [313] -Privacy management: see [102,306,367 [10, 12, 164, 360, 361] -Relaxations of query answering semantics: see [146,147] -Ranking OMQ answers: see [368]…”

On the Complexity of Learning Description Logic Ontologies

“…He constructed an artificial counterexample to prove the result and the argument relies on cryptographic assumptions. Another (artificial) counterexample appears in the work by Ozaki et al (2020) [37]. The argument in this case does not rely on cryptographic assumptions.…”

“…We now highlight some other approaches from the literature for learning DL ontologies, when the focus is on finding how terms of an ontology should relate to each other using the expressivity of the ontology language at hand. These approaches are mainly based on association rule mining, formal concept analysis, inductive logic programming, and neural networks [36].…”