| Capturing and reusing design experience hold great potential for improving design team effectiveness, particularly with the promise of new collaborative Internet design tools. The first step towards leveraging past lessons for design decision-making is to gain access to them. In this thesis, two approaches to accessing and handling design information in teams are taken. One focuses on formal, structured design information and the other on informal, unstructured information.; The role of formal information in design is examined in the adaptation of Quality Function Deployment (QFD), traditionally a co-located design methodology, into a computer-networked process. In this exploration, a software interface was designed to facilitate non-co-located communication among multidisciplinary design teams. The information passed between team members included numerical ratings of product attributes and qualitative rationale of those ratings. While this structured information is essential to QFD, it was noted that much of the informal design discussion that occurs during a co-located meeting is lost in the process.; Informal information is the focus of the second part of this thesis, which deals with the retrieval of design information. Indexing and retrieval of text information are problems that have been studied extensively. However, design information presents an interesting retrieval problem because of the evolving nature of design language. A part may be known as a “thingamajig” at the beginning of a project but eventually becomes a “lead screw” by the end. This evolution can make traditional retrieval challenging. To improve retrieval, thesauri of design terms were manually constructed to augment search. These thesauri included process, component, and function terms. It was found that manually constructed thesauri improved information retrieval from notebooks significantly.; Recognizing the usability difficulties of hand built thesauri, the effectiveness of machine-generated thesauri was also explored. Thesauri were extracted using Singular Value Decomposition. Comparisons of the two approaches to creating thesauri, by hand and by machine, show that manually generated thesauri produce better information retrieval performance. This also includes the trade-off of human overhead required to manually generate the thesauri. However, future work will examine the possibility of improving the effectiveness of machine-generated thesauri through increased document segmentation. |
