2021 IEEE International Systems Conference (SysCon) | 2021
System Level Knowledge Representation for Complexity
Abstract
To develop systems capable of high level cognitive functions such as intelligence, it is necessary to formally capture different types of knowledge, so that they can be used to support complex processes, such as inference and reasoning. The design and engineering of Intelligent Systems to support large distributed socio technical processes increasingly leverages converging techniques from Artificial Intelligence, Knowledge Representation (KR) and Cognitive Architectures. This is resulting in multi layered architectures and AI technologies which one the one hand offer unprecedented capabilities, on the other hand present innumerable, often inconceivable risks. Sophisticated conceptual structures are necessary not only to support the modeling, validation and explanation of complex engineered systems, but primarily to support cognition and conceptualization of the complexities involved, for designers, developers, end users and any stakeholder. Depending on the cognitive makeup of observers, and on the knowledge available, complexity can be conceptualized and traversed following a diversity of methods and patterns. Sometimes complexity can be broken down into cognitively accessible chunks, in other cases however, it cannot be broken down without losing essential information about the system as a whole. Addressing the need to develop cognitive artifacts, methods and techniques that can capture and represent complexity, this paper proposes the outline of conceptual structure that bridges existing approaches which tend to distinguish between cognitive engineering and Knowledge Representation, with the aim to integrate technical and socio technical systems dimensions. The paper presents considerations about cognitive aspects of complex systems theory and practice. It anticipates a convergence between cognitive architectures and KR, introduces the notion of System Level Knowledge Representation and applies it to navigate socio technical complexity in systems engineering. A summary of related work where the System Level Knowledge Representation is being developed and evaluated is also provided.