International audience

The strategic implementation of Artificial Intelligence (AI) technologies in the industry requires extending conventional engineering disciplines to include AI-specific considerations. This allows the management and evaluation of risks associated with AI technologies, thereby unlocking their potential to improve system autonomy. Moreover, this allows to ensure a high level of confidence among stakeholders, such as regulatory authorities and clients. In this context, this paper provides an overview of the findings from the confiance.ai research program, which aims to develop methodological guidelines/ processes for engineering trustworthy AI systems. These processes are the result of collaborative efforts by a large group of experts focused on AI system trustworthiness. Data trustworthiness assessment and risk analysis are examples of these methodological processes.

International audience

The multi-perspective Business Process Modeling (BPM) is grounded upon the idea of constructing distinct and separate models that depict different aspects of the same Business Process (BP). It promotes a comprehensive description of the latter. For an enterprise, such description is crucial for the improvement of its business processes, and hence for remaining competitive. However, none of the modeling languages is endowed with the capability to represent a BP according to all relevant perspectives. A first challenge is then how to improve this capability. At the same time, the multi-perspective BPM is always accompanied with the challenge of managing consistency among the produced models since they are complementary, and consequently share overlapping data. In line with this, this paper presents the proposal of a promising approach addressing these two challenges and based on the mapping between modeling languages.

The Enterprise Knowledge Development (EKD) method promotes a holistic approach in Enterprise Modeling and advocates that a single modeling perspective (product, activity or role) is not sufficient for describing business processes. Instead, it offers a set of modeling formalisms, namely actor-role, role-activity, and business objects formalisms, in order to capture the knowledge on enterprise processes view from different perspectives. On the one hand, this strengthens modeling complex business processes, but on the other hand this sets the requirement for dealing with the consistency among those models describing the same business process according to different perspectives. The systematic literature review we have conducted showed, among its findings, that there is no modeling tool supporting all these formalisms, and hence no way was proposed to handle their consistency. In multi-perspective modeling, the quality of the created models and their capability to be involved in the “design by reuse” depend strongly on the consistency among them. This paper aims to enhance the syntactic and semantic quality of EKD business process models by proposing a set of consistency rules controlling their dependencies. As the definition of semantic consistency rules is far from obvious, we have modeled and investigated a plethora of business processes from different domains. Our proposal is then supported by a tool in order to promote an automated consistency handling and hence to save the time and efforts of process modelers.