On June 30, we held the 6th SCOR online seminar, featuring two guest speakers, Dr. Martin Glauer and Dr. Fabian Neuhaus, both leading researchers in the development of formal ontologies and their applications.
The seminar focused on semantic dependency in ontologies, a concept that is both theoretically profound and practically significant. In the first half of the seminar, Dr. Neuhaus highlighted the problem of semantic incongruity, introduced the notion of semantic dependency, and discussed its various forms. In the second half, Dr. Glauer presented several computational approaches for identifying semantic dependencies in ontologies, including an AI-based method that leverages large language models. He illustrated these approaches through case studies based on existing large-scale ontologies.
The seminar highlighted the strong connection between foundational theory and practical applications in ontology research, reflecting SCOR’s mission. It attracted more than 25 participants and concluded with a lively and engaging discussion.
The abstract, recording, and presentation slides are available below.
To stay informed about upcoming SCOR online seminars and other related events, subscribe to our newsletter using the form at the bottom of the homepage and/or follow us on LinkedIn: https://www.linkedin.com/company/swissonto/.
Martin Glauer & Fabian Neuhaus (Otto von Guericke University Magdeburg)
“Semantic Dependencies in Formal Ontologies”
Abstract
Ontologies often consist of complex logical theories. However, this complexity poses significant challenges during their development. A change to a class or the addition of an axiom may have significant implications for large parts of the ontology. This raises an important question for ontology developers: If we change the semantics of one term by altering an axiom, which other terms in the ontology may be affected and, thus, should be double-checked?
Philosophically, there are strong arguments for semantic holism according to which the semantics of all terms in an ontology are interdependent. However, while semantic holism may be correct, from a practical point of view it is not helpful, because it is not feasible to double-check all terms and all axioms in an ontology each time a minor change is made. Thus, we use semantic dependency as a means to track whether changing the semantics of one term in an ontology may affect the semantics of another. E.g., in SNOMED CT, the class Type 2 diabetes mellitus depends semantically on Endocrine disorder, but not on Procedure declined by patient.
In this talk, we will present different approaches to formalizing a notion of semantic dependency between the nonlogical symbols (e.g., class names and role names) in an ontology. In addition to theoretical considerations, we will also demonstrate the practical use of this approach. To this end, we will present a tool capable of extracting dependencies even from large ontologies and show how these can then be used to identify potential, previously undetected errors in those ontologies.
Short bio
Martin Glauer is a researcher in knowledge representation and neuro-symbolic AI, with a focus on formal ontologies and semantic technologies. In his work he explores the synergies between knowledge representation and machine learning methods, in particular the integration of ontologies into the training process of prediction models in chemistry. He was also the founding developer of the Open Energy Ontology, a domain ontology for energy systems modelling.
Fabian Neuhaus is a researcher at the Otto von Guericke University Magdeburg, where he leads the research group Formale Methoden und Semantik at the Faculty of Computer Science. His work centers on ontology, formal logic, and knowledge representation, with a particular focus on the foundations and applications of formal ontologies in computer science and interdisciplinary research. He has published more than 80 papers on applied ontology, including work on upper-level ontologies and the philosophical foundations of ontological modeling as well as methods for building and maintaining ontologies in information systems, biomedical research, and scientific data integration. He has contributed to various ontologies, including the Basic Formal Ontology, the Relationship Ontology, the Common Anatomy Reference Ontology, the Midlevel Energy Ontology, and the Open Energy Ontology. He contributed to ISO/IEC 24707:2018 Common Logic and the Object Management Group (OMG) Ontology, Model, and Specification Language standards. Fabian Neuhaus is a former Vice President of the International Association for Ontology and its Application (IAOA) and currently serves on its advisory board.