Computer assisted design reflection : a web application to improve early stage product in startup companies

Abstract

This thesis investigates the concept of computer-assisted design reflection. The work details the development of a prototype framework and reflection engine. Reflection is a critical process in design. It allows a designer to learn from experience to improve future results. When applied throughout the design process, it can also improve the quality of design decisions as they are made. We have observed that, unfortunately, design teams in startup environments often do not reflect sufficiently, leading to poor results that can cause a company to fail. A computer-assisted system could prompt reflection regularly and also compile responses from designers and users alike to form a larger dataset for analysis. A web application is uniquely suited to this task because of development simplicity and ease of use for the end user. It is also more prone to adoption by the startup, since many already use a variety of web applications. Through the application of natural language processing and information retrieval theory, the software presented in this research transforms free text inputs into a series of output data clusters that capture both subtly patterns and discrete segmentations in the dataset. Initial testing indicates effective and actionable results from datasets derived from Amazon product reviews. Further testing and development is proposed, beginning with the assembly of large design reflection datasets, as none currently exist. Such datasets would enable system improvements and open up new avenues of research in computer-assisted design.