The notion of the interaction space of an information system

Abstract

Information systems’ (IS) design concerns modeling systems that are dynamic in nature. A dynamic system essentially has two dimensions of concern – static structure and dynamic behavior. The existence of dynamics – or interactions among parts of the system distinguish a dynamic system from a heap or collection of parts. Specification and management of the static aspects of an information system like the data and metadata have been fairly well addressed by existing paradigms. However, an understanding of the dynamic nature of information systems is still low. Currently most paradigms model behavioral properties above an existing structural model, resulting in what may be called “entity centric” modeling. Such a kind of modeling would neglect properties that can be attributed to behavioral processes themselves, and relationships that might exist among such processes. This thesis argues that the dynamics of an information system are best managed by explicitly characterizing an “interaction space” of the information system. An interaction space is defined as an abstract domain that represents the set of all dynamics of the information system. This is contrasted with an “entity space” that represents elements of the static structure of the information system. Recent results on the nature of interactive behavior and of open systems indicate that interaction spaces are characteristically different from the hierarchical nature of algorithmic problem solving. Interaction spaces consist of multiple interactive processes which affect the behavior of one another. Paradigms for the characterization of these spaces are hence explored as part of the thesis.