Supporting example-based ideation and assessment practices in engineering design

摘要

Over the past decade, examples have become the cornerstone of the design process. In essence, designers are no longer developing ideas from scratch, but instead designing through the synthesis of pre-existing design ideas. The repercussion of this new design paradigm is the quality of the design output has become increasingly dependent on the types of examples designers retrieve for inspiration. Although examples use is pervasive in many design disciplines, little research has been conducted on how and why designers use examples during the design process. This is problematic for the understanding of design activities, as well as for the development of more effective design tools. In addition, example use has not been studied outside of the idea-generation process and thus little is known about what other uses examples could have. For example, one???s ability to identify high quality design examples during the design process has been linked to the designer???s expertise, yet no study to date has explored how examples could be used as an assessment technique. Therefore, the purpose of this dissertation was two-fold. First, we sought to provide an understanding of how and why examples are used in design practice and how we can better support these methods through the enhancement of existing design tools or the creation of new ones. Next, we sought to understand how ratings of example quality can be used to predict one???s design expertise in evaluating ideas, creating ideas and critiquing ideas. In total, six experiments were conducted. The results revealed several challenges to supporting example usage in design. First, the types of examples collected for inspiration can have a negative impact on the designer???s ability to develop innovative solutions. As such, new computational tools are needed that help individuals collect a diverse example set and develop new design directions. In addition, assessments of student design competence are based primarily on subjective evaluations of student performance by the course instructor, a design expert. However, these experts are subject to cognitive biases based on their own beliefs and expectations. Our example-based assessment method utilizes the Bayesian Truth Serum algorithm which removes instructor bias by taking them out of the evaluation criterion. Our results show a positive correlation with student???s ability to analyze and evaluate design ideas, but not their ability to generate innovative solutions. Implications and future research directions are discussed.