Supporting the Design of Socio-Technical Systems by Exploring and Evaluating Design Alternatives
Volha Bryl · 2009
A challenging aspect of modern information systems is the growing involvement of humans and organizations in system structure and operation. Organizational environment in which software operates, the software system itself, the related hardware components and human users are often interdependent in a non trivial way, so that it is problematic to define a system boundary. We argue that an interdisciplinary notion of a Socio-Technical System (STS) is the one that captures the above mentioned aspects. Unlike traditional computer-based system, socio-technical systems include in their architecture and operation not only software and hardware components, but also organizational and human actors. Such systems are regulated and constrained by internal organizational rules, external laws and regulations. Among the challenging problems related to the analysis and design of a STS is the problem of understanding the requirements of its software component and the way in which the structure of human and organizational activities is influenced by introducing technology. In particular, an important element in the design of a socio-technical system is the identification of a set of dependencies among actors which, if respected by all parties, will fulfill all stakeholder goals, namely, the requirements of the STS. In this thesis, we present a framework which aims at supporting the design of sociotechnical systems, specifically the design of a network of inter-actor dependencies intended to fulfill a set of initial goals. The support comes in the form of a structured process and a tool that is founded on an off-the-shelf AI (Artificial Intelligence) planner which generates and evaluates alternative assignments of actor dependencies to identify an optimal or good enough design. We explore a range of measures for evaluating optimality, inspired by AI planning, multi-agent systems, social networks and Economics. We report on the application of the framework to the domains of secure systems design, safety critical systems, dynamic reconfiguration of an STS, as well as to the problem of instantiation of STS designs. We are also experimenting with our prototype tool to evaluate its scalability to realistic design problems.