Page 12 - Sirius_Annual_Report_2021

Ontology Engineering
The digital transformation of the industry depends on rich information models in order to support automation of specialized and knowledge intensive tasks. These models must be intelligible and usable by both computers and humans and should ideally represent the concepts and relationships in a manner to which domain experts are accustomed. This way users and systems may explore and extract implicit information from data through the help of automated reasoning without the need for understanding the technical details of how and where the data is stored.
Martin G. Skjæveland
However, the construction, maintenance, and use of such a model, called an ontology, are far from straight forward. Creating and maintaining
a high-quality ontology requires close collaboration between domain experts, information modellers, and ontology experts to ensure that the model works as intended. Furthermore, an ontology quickly becomes
a very complex artefact in
The pattern-based ontology engineering project has developed the Reasonable Ontology Templates (OTTR) framework. OTTR is a language and framework for repre- senting and instantiating recurring patterns for engineering ontologies. This allows building and interfacing with the ontology at a higher level of abstraction than what is possible using the current standard ontology language OWL. This includes:
• Building ontologies and knowledge bases by instantiating templates;
• presenting, transferring and visualising the knowledge base as a set of template instances at different levels of abstraction; and
• securing and improving the quality and sustainability of the knowledge base via structural and semantic analysis of the templates used to construct the knowledge base.
Members of the project are Martin G. Skjæveland, Leif Harald Karlsen, Christian Kindermann, Oliver Stahl. The framework is available as open source specifications and applications which are in active use by several industrial partners in and outside of SIRIUS, including DNV GL, Aibel, Grundfos and CapGemini.
For more information about OTTR, including interactive examples, specifications and research papers, see its homepage http:/ottr.xyz.
In addition to the OTTR framework, the project works on research topics for identifying and characterising patterns in ontologies. OWL ontologies are built and maintained on the basis of all sorts of methods and methodologies using a wide range of tools. As ontologies are primarily published as sets of axioms, their underlying design principles often
order to express and make use of all the desired information objects. This makes maintaining the ontology a real issue.
The aim of the ontology engineering research program is to develop tools and methods that improve the usability, efficiency and quality of ontology development, main- tenance and use in the industry, by
• lowering the barrier for domain experts to understand, build, and use ontologies without the support of ontology experts.
• providing programmers and information modellers with powerful interfaces for interacting with and exploiting the knowledge captured in the ontology with existing software platforms.
• equipping ontology experts with powerful tools to oversee the development of the ontology
Work in the research program is primarily performed in two projects: the pattern-based ontology engineering project and the information modelling framework project.
12 | SIRIUS ANNUAL REPORT 2021