Effects of Data Exploration and Use of Data Mining Tools to Extract Knowledge from Databases (KDD) in Early Stages of the Engineering Design Process (EDP

This thesis describes original research work where the objective was to provide teams with access to data, and observe the effect of its use at the early creative stages of the engineering design process. Following a theoretical research on the use of information technologies to support idea generation, and the use of data as creative input, a procedure was designed following the Knowledge Discovery from Databases process, and tried over several iterations of improvement working with creative teams in different contexts.;After two exploratory studies, three cases were performed where the researcher attempted to better support the different stages of the EDP through the application of data from patent mining. To observe the differences, we provided three levels of access to explore data in a data mining tool: low, intermediate and high.;Case 1 - Participants in a creativity session were asked to identify needs or problems (first stage of the engineering design process). They were given intermediate access to explore data in a data mining tool, meaning they could explore, but not make new searches or add data. The analysis of the results indicates that participants gravitated towards terms and keywords related to previously generated ideas, thus the increase in novelty was low. In order to correct the issue of intermediate exploration, it was decided to train participants in the use of the data mining tool for subsequent cases; if teams have more freedom to explore data, they can potentially generate more novel combinations.;Case 2 - Teams tasked with engineering challenges in a course were trained in the use of the data exploration tool. They were then invited to continue using the tool to generate new ideas. In this case, teams had high access to the data exploration tool; they were able to add data, and make searches. Teams who chose to explore data for creative support found improvements or components from existing solutions to advance their own design, and received more positive evaluations by a jury of experts. However, the objective of obtaining more diverse or novel solutions was not achieved. A possible explanation is that the use of the tool can overwhelm participants with too many options to explore, leading teams to return to known solutions. A possible counteraction to resolve the issue of too many options is to have an external actor (such as a moderator) extract keywords from the data, and provide participants with these terms to combine into novel ideas.;Case 3 - Teams participating in an innovation contest were given keywords selected by an expert on the tool. In other words, participants had low access to explore data in a data mining tool. The researcher performed the data analysis for two challenges in the competition, and selected keywords relevant to the knowledge base of the problem. The results show that teams who selected the keyword supported challenges generated more diverse and novel ideas, compared to teams without the support. By providing relevant keywords, it was possible to obtain the benefits of the KDD without the issues of training participants on the use of the tool, and the resources teams would have to dedicate to explore the data.;It was concluded that data and KDD can be used as a creative input for an EDP at different stages. It is recommended to determine whether the objective of including data in an EDP effort is to generate a novel idea or to solve a problem. To generate novel ideas, it seems preferable to provide data in the form of keywords selected by an external actor, to prompt original combinations. If the team is searching for incremental improvements or elements of existing solutions, then it appears to be beneficial to have access to a knowledge base to explore. It is important to delimit the exploration to avoid becoming stunned because of the amount of available information.;For the three experiences, the software IPMetrix was used to perform the data mining. The process of data selection, loading, cleaning and transformation is described in each chapter, according to the work performed on the data for the specific case.