Increasing calls for a transition towards a more sustainable economy have led to sustainability-oriented innovations (SOIs). For the development of SOIs, various technologies are emerging that go along with, inter alia, high uncertainties but might have the potential to transform an existing industry. Thus, identifying and evaluating emerging sustainability-oriented technologies is crucial both for companies to evaluate business opportunities and for researchers and policy makers to support the transition towards a more sustainable economy.
Applying semantic similarity analysis and unsupervised machine learning algorithms on scientific research articles in the field of precision agriculture (PA); a case example where several SOIs occur, we create topic clusters to evaluate fast growth and coherence of technologies within PA. Here, we compare two different approaches. Namely, we use spectral clustering and hierarchical clustering. To evaluate, analyse and compare the clusters, we apply different cluster labelling approaches and analyse the evolution of the clusters over time. Our results show that spectral clustering seems to outperform hierarchical clustering in the field of PA.
The presented research was conducted in cooperation with the Fraunhofer Institute for Technological Trend Analysis INT. As co-researchers we would like to name Dr. Michael Wustmans, Dr. Marcus John and Prof. Dr. Stefanie Bröring and thank them.