An agent-based approach to the governance of complex cyber-infrastructures

With the increasing digitization of our society, more of our daily life is being affected by software systems supporting us taking (partially) automating decisions (about visa applications, job applications, credit placements, mortgages, insurances, …), by means of rule-based, data-driven, or hybrid methods. This transformation motivates the introduction of engineering approaches to govern the design, development, deployment, and maintenance of digital systems in alignment with regulations. This dissertation tackles this challenge by focusing on computational agents, and their interaction with norms, to conceive and develop a set of design and modelling tools, computational methodologies, and insights concerning the governance of socio-technical systems. Agent models can be utilized to capture the different stakeholders participating in the social system, each with their own type of beliefs, intentions, preferences. Norm models sets expectations and thus criteria for the evaluation of behaviors; as such, they form a fundamental basis for social coordination mechanisms and for judgments about compliance.
This dissertation focuses on agent modelling and the associated model-execution as a principal step in system design and policy-making activities on digital infrastructures. It introduces an agent-based programming framework (named ASC2) based on the belief-desire-intention (BDI), alongside a scalable multi-agent system as runtime environment. The dissertation further elaborates on the value of utilizing mainstream software development tools in agent-based programming and illustrates this by bridging ASC2 with multiple tools such as testing, debugging and even DevOps tools.