Thomas PaulsenText Mining in der Szenarioanalyse
Eine methodische Erweiterung zur Reduktion des Aufwands und der Subjektivität bei der Sammlung von Einflussfaktoren
QM – Quantitative Methoden in Forschung und Praxis, volume 50
Hamburg 2020, 346 pages
ISBN 978-3-339-11544-7 (print) |ISBN 978-3-339-11545-4 (eBook)
About this book deutschenglish
Scenario analysis has been one of the most important instruments of strategic planning since its use in the Shell Group in the 1970s. The aim of scenario analysis is to create multiple alternative images of the future, which can be used as tools for decision-making. All methods of scenario analysis start with the collection of factors that influence the future of the object of investigation. These are called influence factors. However, the classical collection of influence factors is very time consuming and can be biased by the opinion of individuals. These characteristics lead to the fact that carrying out scenario analyses is very expensive and that the results are not always reliable.
This dissertation aims to overcome this criticism. To this end, the data obtained will be analyzed in an innovative, computer-aided way. Specifically topic models, a group of models originating from the area of text mining, are used. The goal of topic models is to search automatically for semantically coherent word groups (topics). This search takes place without the inclusion of prior knowledge. For the course of the dissertation topics are equated with influence factors. The use of topic models for the collection of influence factors is investigated in the context of a real scenario project for the city of Münster. In this case study, the effort for the analysis of the data was reduced to 20%. Furthermore, it could be shown that the quality of the influence factors is comparable to the ones generated classically. The described procedure increases the efficiency of scenario creation and thus contributes to the dissemination of this instrument. From a scientific perspective, a strategy was presented that allows assessing the quality of the influence factors. This strategy can be transferred to a variety of future scenario projects.
Keywords
AufwandsreduktionBewertung von Topic ModellenEffizienzsteigerungLatent Dirichlet AllocationMachine LearningStrategische PlanungStrategisches ManagementSzenarioanalyseText MiningTopic ModelingTopic ModelleZukunftsforschungIhr Werk im Verlag Dr. Kovač
Möchten Sie Ihre wissenschaftliche Arbeit publizieren? Erfahren Sie mehr über unsere günstigen Konditionen und unseren Service für Autorinnen und Autoren.