Knowledge modelling is a cross-disciplinary field that focuses on how to capture, preserve, and apply existing knowledge.

Within the field of knowledge modelling, frequently used terms are knowledge graphs, visual methods for representing knowledge, and ontologies (or ontology engineering). Ontologies, much like knowledge graphs, represent a (primarily visual) way of representing the fundamental concepts of importance when representing a certain type of knowledge, as well as how these concepts relate to each other.

Although there are different approaches to knowledge modelling, the end-goal of a knowledge model remains consistent. Knowledge models turn something abstract and untouchable (knowledge) into something tangible, which can be examined and distributed. Knowledge models are additionally of relevance for Artificial Intelligence. Through knowledge modelling, it is possible to capture the knowledge of experts, which the AI can then use and apply for its own internal workings, or it can be used to create an internal representation of the knowledge the AI itself has. The latter application of knowledge modelling can be useful for the creation of more transparent and explainable AI.

Bayesian networks

A Bayesian network is a probabilistic graphical model that can be used to describe (causal) relations between random variables in complex technical systems. Queries about the system can be answered with inference techniques that are grounded in probability theory. Expert knowledge about the system components and interaction of these components can be used to define the nodes of the graph which represent random variables. The links between the nodes represent conditional probabilities that can be learned from measurements of the system and its components. Because of their graphical structure and the clear relation of the vertices and edges with expert knowledge, Bayesian networks are more transparent and can be more easily be interpreted than neural networks in which the nodes and links do not have a direct connection with expert knowledge.

What does TNO offer on knowledge modelling?

When it comes to Bayesian networks, TNO offers a systematic approach for the generation of Bayesian networks based on existing system specifications. TNO furthermore uses the Bayesian network approach for the diagnostics and root-cause analysis of complex technical systems.