Proceedings of the

The 33rd European Safety and Reliability Conference (ESREL 2023)
3 – 8 September 2023, Southampton, UK

Computational Ontologies: Discussion of a Research Protocol in Human Sciences on the Organization of Industrial Risk Management

Emmanuel Plot1,a, Maria Chiara Leva2, Micaela Demichela3, Vassishtasaï Ramany B.P.4, Frederic Baudequin1,b and Marine Boutillon1,c

1Ineris, France.

2Technological Université Dublin, Ireland.

3Politecnico di Torino, Italy.

4SNOI, France.


Industrial risk management in applied setting is more and more faced with the need to provide a unified conceptual picture favourable to the elaboration of risk assessments and risk monitoring approaches, while at the same time accommodate the use of a plurality of data, knowledge, models and expertise that come from different areas, stem from different point of view (field operator vs designers), and reflects also different belief. A central difficulty would be to find bridges between the different levels of abstraction and conceptualization that experts may use, manipulating notions that are only apparently common. It is a problem of conceptualization, because it is the concepts that make it possible to organize the representations, to manipulate the data, the methods, the models and to coordinate the contributions of the different experts. Here, the use of digital technology is essential, as databases allow more easily than paper (or pdf files) to embrace diversity, heterogeneity, and dynamic interactions of knowledge. The question is how to design the right application or platform to help solve this problem, and also to help deliver a common operational pictures that can be operationally deployed. What are the underpinning concepts that can deliver such a purpose. This paper presents the results of a research, started in the framework of the European project Tosca (2013), on the development of ontologies as a support for the safety management systems (cf. Seveso regulation). The thesis that we wish to present in this article is that computer language is indispensable for the design of concepts useful for risk management, as soon as these risks are "major" and their management therefore requires an important level of precision. Computer tools should be considered as a research protocol in human sciences, not only as a support to instantiate concepts that could be elaborated before their computerization. In other words, IT is essential for thinking about major risk management organizations.

Keywords: Complexity, Cognitive bias, Risk assessment, Risk management, Dialogue, Digitization.

Download PDF