Judging the similarity of part shapes
Introduction
The ongoing advances in Machine Learning and Artificial Intelligence will have an impact on the way new apps, devices, and systems will be designed and implemented, and arguably this will accelerate the pace of automation itself. From static interfaces and systems, we will increasingly move to adaptive interfaces and systems. Currently, systems are adaptive by either 1) adjusting the interface components when interacting with a computer screen or a mobile screen, or 2) studying retrospectively the user's use patterns to anticipate potential needs through customised recommendations. However, research is still sparse regarding how interactive intelligent systems (for future educational & work activities) will be able to detect the human variability – i.e. user expertise and needs and adapt the adaptive system and interface accordingly.
This exploratory research aimed to address this gap.
Critically, this research had the objective to investigate how participants with a background and/or experience in mechanical engineering and product design would perceive and cluster given shapes, and use this information to inform the training of the DCS Machine Learning Tool. This study involved an experiment through an online survey, with the overarching aim to inform Machine Learning on how to retrieve similar parts from a big dataset of shapes for use in engineering design. In one of the experimental tasks, the participants were asked to cluster a random collection of part shapes based on their own reasons, and to describe in words the criteria they used to group the shapes (as shown in the picture below).
The responses showed differences between novice engineers and expert engineers. The novices (95% of respondents) grouped the displayed objects by shape similarity; on the contrary, experts provided more complex solutions (e.g. they clustered the objects by manufacturing process, by functionality, etc.) (98% of respondents).
These exploratory results suggest consideration of the differences between experts and novices in terms of how they make sense of their work environments, and address/solve specific problems, and therefore the role played by user expertise when using future intelligent interactive systems.