project summary: 

The Challenge: Lost Knowledge in Architecture 

Architectural design generates enormous amounts of valuable knowledge, yet most of it is ephemeral. It lives in the memories of team members, in meeting conversations, and invisibly inside complex digital models, and it is largely lost when project teams disband. The result is systemic knowledge loss across the Architecture, Engineering, and Construction (AEC) sector: firms repeatedly “reinvent the wheel,” efficiency suffers, and opportunities for innovation are missed (Carrillo et al., 2000; Kamara et al., 2003). 

The problem is becoming urgent. A large share of Australia’s construction workforce is approaching retirement, while only a small minority of organisations systematically capture the knowledge of departing staff. Post-project reviews, where they happen at all, rarely capture knowledge in a reusable form. 

This research addresses a critical problem: the inability of architectural organisations to systematically learn from their own historical project data to improve future decision-making. 

What We Have Found: Beyond the “BIM Utopia” 

The first phase of the research, a Systematic Literature Review (SLR) of 31 peer-reviewed studies spanning 2000–2025 that is currently under review at the Journal of Architectural Engineering, confirmed and sharpened the project’s two founding knowledge gaps: 

• Knowledge is dynamic, but capture methods are static. Design decisions are proposed, debated, revised, and superseded over weeks and months, yet no reviewed method tracks this temporal evolution. Existing approaches produce snapshots of a moving process. 

• Data and process remain disconnected. Technology-driven solutions, particularly Building Information Modelling (BIM), record what was designed but not how or why. The reasoning, negotiation, and trade-offs behind decisions, the sociotechnical reality that “BIM utopia” visions overlook (Miettinen & Paavola, 2014), remain fragmented across people and time. 

The review also revealed a structural imbalance in the field. The large majority of the knowledge capture methods identified target technical artefacts only, with only a small fraction addressing the social dynamics of design. Manual methods capture rich context but do not scale; automated methods scale but miss tacit reasoning. Strikingly, no reviewed study demonstrates empirical evidence of successful cross-project knowledge transfer. 

The Approach: A Temporal Knowledge Graph of the Design Process 

Building on Sociotechnical Systems theory, which emerged from foundational postwar studies of industrial work (Trist & Bamforth, 1951; Trist, 1981) and has recently been applied to design practice (Pirzadeh et al., 2021), and on a bottom-up, “no-model” view of knowledge (El-Diraby, 2023), the research is developing a computational knowledge capture framework that models the design process as an evolving sociotechnical network. 

The key insight is that design knowledge already exists in the digital traces firms routinely produce: BIM model histories, shared-platform activity, meeting records, and project correspondence. It simply isn’t structured, connected, or retrievable. The proposed framework passively structures these traces into a temporal knowledge graph in which people, artefacts, and decisions coexist as interconnected nodes, with every relationship carrying temporal information so the evolution of knowledge, including superseded and revisited decisions, is preserved rather than overwritten. Recent advances in AI-based knowledge extraction and hybrid graph retrieval make it possible to build such a graph from unstructured project data and to query it in natural language: “Why did the facade material change?” 

The goal is to enable firms to achieve genuine ‘double-loop learning’ (Argyris & Schön, 1996): not just correcting errors, but questioning and improving the standards and processes that guide future projects. 

Where the Research Is Now 

The project is structured in three stages, and has moved from theory into empirical work: 

1. Understanding the process (complete / under review): The SLR is complete and under peer review. Exploratory case studies of completed Australian architectural projects (2015–2025) are underway, tracing how actors, artefacts, and decisions actually interact across the design phases, complemented by a practitioner survey conducted globally and through the Arch_Manu industry partner network. 

2. Preparing project archives (in progress): A de-identification and pre-processing pipeline that transforms raw project archives (models, correspondence, meeting records, and shared-platform data) at scale into clean, privacy-safe, analysis-ready form, preparing them for ingestion into the temporal knowledge graph. 

3. Building and validating the framework (commencing): The temporal knowledge graph framework is being developed and will be validated through retrospective analysis of completed projects and evaluation by a panel of industry and academic experts, with data sovereignty, privacy, and ethical safeguards built in from the outset. 

Anticipated Impact 

• Mitigate systemic knowledge loss as teams and experienced staff transition between projects and out of the profession. 

• Support evidence-based decision-making, giving less experienced staff access to the reasoning behind past decisions, not just their outcomes. 

• Foster continuous organisational learning, turning project archives from unsearchable storage into a living organisational asset. 

Ultimately, this research seeks to transform how architectural project knowledge is captured and applied, enhancing operational efficiency, improving design outcomes, and offering the AEC sector a new, more realistic model of design as a complex sociotechnical system. 

Key References 

Argyris, C., & Schön, D. (1996). Organizational Learning II: Theory, Method and Practice. Reading, MA: Addison-Wesley. 

Carrillo, P. M., Anumba, C. J., & Kamara, D. J. (2000). Knowledge Management Strategy for Construction: Key I.T. and Contextual Issues. Proceedings of the International Conference on Construction IT, Reykjavik, Iceland, 155–165. 

El-Diraby, T. E. (2023). How typical is your project? The need for a no-model approach for information management in AEC. Journal of Information Technology in Construction, 28, 19–38.  

Kamara, J. M., Anumba, C. J., & Carrillo, P. M. (2003). Conceptual framework for live capture and reuse of project knowledge. Proc., CIB W78’s 20th Int. Conf. on Information Technology for Construction, Rotterdam, Netherlands, 178–185. 

Miettinen, R., & Paavola, S. (2014). Beyond the BIM utopia: Approaches to the development and implementation of building information modeling. Automation in Construction, 43, 84–91. 

Pirzadeh, P., Lingard, H., & Blismas, N. (2021). Design Decisions and Interactions: A Sociotechnical Network Perspective. Journal of Construction Engineering and Management, 147(10), 04021110. 

Trist, E. L. (1981). The evolution of socio-technical systems: A conceptual framework and an action research program. In A. H. Van de Ven & W. F. Joyce (Eds.), Perspectives on organization design and behavior (pp. 19–75). Wiley. 

Trist, E. L., & Bamforth, K. W. (1951). Some social and psychological consequences of the longwall method of coal-getting. Human Relations, 4(1), 3–38.