Abstract
Organizations worldwide are supporting their processes and decisions with enterprise systems (ES). Large amounts of data are produced and reproduced in these increasingly complex sociotechnical systems, opening new opportunities for the adoption of self-supervised learning techniques. Complex networks are viable solutions to create models that learn from data. This chapter presents (1) a review on the possibilities of networks for self-supervised learning, (2) three cases illustrating the potential of complex networks to address the autopoietic nature of ES (adoption of enterprise resource planning, web portal development, and healthcare data analytics), and (3) a framework to mine sociotechnical patters uncovering the entanglement of human practice and information technologies. For theory, this chapter explains the potential of complex networks to assess enterprise systems dynamics. For practice, the proposed framework can assist managers in establishing a strategy to continuously learn from their data to support decision-making in self-adapting scenarios.
TopMethod
Design science research (DSR) has its roots in the sciences of the artificial (Simon, 1996), aiming to simultaneously design innovative artifacts and contribute to scientific advances (Hevner, March, Park, & Ram, 2004). An information systems artifact may “refers to a system, itself consisting of subsystems that are (1) a technology artifact, (2) an information artifact and (3) a social artifact, where the whole (the IS artifact) is greater than the sum of its parts (the three constituent artifacts as subsystems), where the IT artifact (if one exists at all) does not necessarily predominate in considerations of design and where the IS itself is something that people create” (Lee, Thomas, & Baskerville, 2015). Theory and artifact production must be balanced during the research lifecycle (Baskerville, Baiyere, Gregor, Hevner, & Rossi, 2018; Deng & Ji, 2018).
Key Terms in this Chapter
Enterprise Systems: Software packages that support organizational processes and services, generating valuable data to support decision-making. These systems can solve enterprise-wide problems and improve data integration.
AiLLe: A four step framework to model and visualize complex networks in a cybernetic environment where humans and machines are significantly interdependent.
Complex Adaptive Systems: A particular form of complex system composed by multiple interactive elements that reveals the capacity to change and learn from experience.
Complex Network: A graph of connected nodes that are used to represent real systems. The nodes interact in multiple different ways producing network configurations that may evolve over time.
Information Entropy: An important measurement of uncertainty in information theory that was introduced by Claude Shannon in 1948. The concept is similar to thermodynamics entropy that reveals the possible outcomes of a specific variable.