[Paper Review] An Applied Knowledge Framework to Study Complex Systems

Juste Raimbault|arXiv (Cornell University)|Jun 22, 2017

Business Strategy and Innovation10 references3 citations

TL;DR

This paper proposes an applied knowledge framework that models knowledge production on complex systems as co-evolving perspectives within interdependent knowledge domains, using mixed-methods analysis of a geographical theory of urban complexity. The framework enables reflexive, interdisciplinary management of complexity by integrating epistemological structure with practical application in system design and modeling.

ABSTRACT

The complexity of knowledge production on complex systems is well-known, but there still lacks knowledge framework that would both account for a certain structure of knowledge production at an epistemological level and be directly applicable to the study and management of complex systems. We set a basis for such a framework, by first analyzing in detail a case study of the construction of a geographical theory of complex territorial systems, through mixed methods, namely qualitative interview analysis and quantitative citation network analysis. We can therethrough inductively build a framework that considers knowledge entreprises as perspectives, with co-evolving components within complementary knowledge domains. We finally discuss potential applications and developments.

Motivation & Objective

• To address the lack of knowledge frameworks that simultaneously account for epistemological structure and practical application in complex systems research.
• To develop a framework that reflects the co-evolution of knowledge domains and their carriers in scientific knowledge production.
• To provide a tool for managing complexity by making implicit methodological and epistemological choices explicit in system modeling and design.
• To demonstrate the framework's generality through case studies in geography and engineering, showing transferability across domains.
• To lay the groundwork for formalizing the framework in future work, particularly through integration of dataflow machines and ontologies.

Proposed method

• Conducts mixed-methods analysis of knowledge production in a geographical theory of complex urban systems, combining semi-directed interviews with key researchers and quantitative citation network analysis.
• Identifies and maps knowledge domains as interdependent components that co-evolve through feedback and interaction, using a perspective-based approach.
• Applies a cognitive epistemology rooted in perspectivism, where each perspective is a structured ensemble of knowledge domains with shared purpose and carrier.
• Uses the concept of weak emergence to define hierarchical relationships between ontology elements within a perspective, enabling canonical decomposition.
• Projects knowledge domains as a complete network to explore all possible binary relations (e.g., constraints, knowledge transfer, methodological synthesis), illustrating co-evolutionary dynamics.
• Proposes a path toward formalization by coupling a dataflow machine model (from [8]) with an ontology-based representation (from [11]), linking data, modeling, and empirical/theoretical components.

Experimental results

Research questions

• RQ1 How can knowledge production on complex systems be structured epistemologically while remaining directly applicable to system design and management?
• RQ2 What role do co-evolving knowledge domains play in the development of scientific theories on complex systems?
• RQ3 How can mixed-methods approaches—qualitative interviews and quantitative citation analysis—reveal the epistemological structure of knowledge production?
• RQ4 In what ways can a perspective-based framework support reflexivity and integration across disciplines in complex systems research?
• RQ5 What formal foundations could underlie a scalable and systematic application of the knowledge framework in modeling and simulation?

Key findings

• The construction of the Evolutive Urban Theory was driven by co-evolution among knowledge domains such as modeling, empirical data, and theoretical frameworks, rather than linear progression.
• Interviews with key researchers revealed that methodological choices were reflexive and context-dependent, with no single domain dominating the knowledge production process.
• Citation network analysis showed that knowledge domains were interlinked through feedback loops, with some domains acting as catalysts for cross-domain integration.
• The framework’s projection as a complete network of domains revealed diverse relation types—constraints, knowledge transfer, and methodological synthesis—indicating dynamic co-evolution.
• The ontology-based decomposition of perspectives using weak emergence provides a canonical structure for perspectives, enabling systematic analysis of knowledge components.
• A formal path is proposed to integrate dataflow machines and ontologies, suggesting a foundation for future mathematical formalization of the framework.