An Introduction to Knowledge Representation and Ontology Development for Systems Engineers

Effective communication requires a common vocabulary. An ontology provides a description of the terminology, concepts and relationships for a particular area of interest. An ontology may be viewed as a declarative encoding of the meaning of the domain vocabulary terms, thus making it a key to enabling communication. For systems that are used by people whose understanding of a domain is not necessarily consistent, an explicit description of the important terms can be extremely useful. Many commercial companies have successfully deployed applications with increasing use of semantics such as taxonomy-based search and navigation services. Rule-based manufacturing, product configuration, and financial services systems have been relatively common in those industries for many years. Fewer organizations have successfully deployed semantically rich systems that incorporate ontology-based metadata, sophisticated reasoning and explanation support. The technology has been around for decades, though its use for web-based applications is relatively recent, and it remains difficult for some people to understand, let alone use effectively. This tutorial provides an overview of the knowledge representation landscape and attempts to demystify some of the “black art” of ontology development. We will outline basic methodology steps developed from a combination of Domain analysis methodology from software engineering IDEF methods developed for the US Department of Defense Best practices developed through extensive experience and lessons learned, with a focus on problems in software and systems engineering Examples from systems engineering will be provided, with emphasis on ontology development in UML using the Ontology Definition Metamodel, applications that lend themselves to vocabulary development in the Web Ontology Language (OWL), and use of the these technologies together with models developed using the OMG's Systems Modeling Language (SysML).