University of California, Berkeley
Berkeley, CA 94720-1740
Professor Andy Dong is an alumnus of the University of California, Berkeley, having graduated with High Honors in mechanical engineering (BS ’92) and receiving both his MS and PhD from Berkeley (MS ’95, PhD ‘97). His research aims to understand the sources of competitive advantage in engineering design through the discovery of knowledge structures underlying design processes and objects of design. His research theorizes that the structure of design knowledge has real effects, such as the productivity of design or the progress potential of a new product. Through a unified mathematical perspective grounded in spectral analysis and the cognitive aspects of design thinking, he has discovered structural regularities in engineering processes and products which have real effects on their performance, productivity, progress potential, and robustness and resilience.
To view Professor Dong's CV, please click here.
The principal purpose of my research is to discover the significance of the structuring of knowledge for an understanding of the formation and innovation of designed objects. Discovering the underlying structure of a set of entities is a fundamental challenge in science. Some of the most pivotal discoveries have been about structural forms, such as the double-helix structure of DNA. Until the double helix structure was discovered, we really didn't know how DNA worked. Similarly, while engineering research has been overwhelmingly preoccupied with representations of engineering knowledge - equations, diagrams, domain-specific models - the field was not asking whether there were underlying patterns or structures of engineering knowledge regardless of its representation in an equation or a diagram or natural language. Engineering knowledge remain locked within the specificities of the problem under consideration.
In my research, I am simultaneously investigating the structure of knowledge, and by generating insights into the structure, solving important engineering design problems.
Across a number of studies, I have shown that the structure of design knowledge can be revealed through statistical language processing of patterns of word co-occurrence and semantic links (DOI: 10.1016/j.destud.2004.10.003 and DOI: 10.1017/S0890060406060033), the patterns of symbolic co-occurrence in mathematical models of engineered systems (DOI: 10.1017/S0890060409990175), and in the patterns of component inter-relations (DOI: 10.1115/1.4025490). Through a unified mathematical perspective grounded in spectral analysis (eigenvalue and singular values of graphs), I have discovered that the structure of design knowledge has real effects, including the productivity of design, progress potential of a new technology, and the robustness and resilience of the engineered product or system. Two hypotheses I have confirmed have led to important changes and advances.
Productivity Hypothesis: The structure of knowledge produced by productive design teams differs from the structure of knowledge produced by ineffective teams.
Where the orthodoxy had been that words themselves embody design knowledge, my research showed the surprising result that design knowledge resides in the distributed relations between words. The dimensionality reduction method of latent semantic analysis described in this article (DOI: 10.1016/j.destud.2004.10.003) has become increasingly adopted in social science as a means to understand small-group behaviour.
Innovation Hypothesis: The structure of knowledge has affordances, which affects the rate of progress of technological innovations.
I confirmed this hypothesis by matching the knowledge complexity analyses of energy harvesting products, such as solar and wind, with observed learning rates (DOI: 10.1016/j.techfore.2014.02.009). In doing so, I made an important correction and contribution to studies aiming to forecast the progress potential of products: the fundamental factor in the progress potential of a product is the degree of complexity of the knowledge associated with the product's core technology and the configuration of the parts and sub-systems around the core technology. This complexity produces variance in relation to the amount of effort necessary for technological innovation.