Machine learning in the front end of innovation : trying to identify promising ideas for new product development

This thesis explores several machine learning approaches in the field of natural language processing and tests them for applicability in the front end of innovation. Ideas are the basis for successful innovations. There is a vast number of smart ideas around waiting to be identified. Certain machine learning tools have the potential to support innovators in the identification of promising ideas by analysing extensive amounts of text data and derive meaning from it. In the innovation context, ideas are strongly related to needs. Hence innovation can be interpreted as a matching of needs and solution or as a creative solving of problems. The four different experiments in this research refer to these theoretical concepts. The input data for the machine learning tools is retrieved from an online idea contest calling for innovative product ideas that reduce the plastic waste, and reports by renown organizations and scientists about global plastic pollution. The ideas represent the solution landscape, whereas the reports refer to the need landscape in the area of plastic pollution. The evaluation of the results is mainly guided by the applicability in practice. It reveals rewarding benefits, as well as single shortcomings of the usage of machine learning approaches at the beginning stage of innovation. Moreover, in the course of the thesis the relationship between humans and machines receives particular attention.submitted by Julian WahlUniversity of Innsbruck, Masterarbeit, 2019(VLID)363355