Abstract
Generating new business concepts is an important part of founding new start-ups as well as innovation in existing companies. We identify rising and falling trends in different domains, via topic modelling of text data published over time. Topic modelling is an important tool to classify and cluster documents. Based on the top2vec method, which uses common embeddings of documents and words to not only find but also describe clusters, we have implemented an incremental variant, tracing of growth and decline of these clusters over time. Identification of these trends over large text data collections enables a decision support for innovators, by identifying rising trends, and declining opportunities in different domains. The method was tested and evaluated on the example of arXiv articles. Visualizations of the clusters and descriptions serve to provide people with an interface to identify the trends. In the future, this method can build a foundation of decision support systems that generate innovation ideas based on upcoming trends in research.
Originalsprache | Englisch |
---|---|
DOIs | |
Publikationsstatus | Veröffentlicht - Sep. 2022 |
Veranstaltung | 21st International Conference on Modeling & Applied Simulation, 18th International Multidisciplinary Modeling & Simulation Multiconference - Rom, Italien Dauer: 19 Sep. 2022 → 21 Sep. 2022 https://www.msc-les.org/i3m2022/ |
Konferenz
Konferenz | 21st International Conference on Modeling & Applied Simulation, 18th International Multidisciplinary Modeling & Simulation Multiconference |
---|---|
Land/Gebiet | Italien |
Ort | Rom |
Zeitraum | 19.09.2022 → 21.09.2022 |
Internetadresse |