Digital business foresight: Keyword-based analysis and CorEx topic modeling

Abstract

While previous research develops big data tools to identify weak signals of technological and business changes, we lack an understanding of how to link weak signals with new business concepts and use them to evaluate performance of startups in an automated way. To address this research gap, we analyze the business description of digital companies extracted from the CrunchBase metadata and use a keyword-based text mining approach to collect business-related weak and strong signals for the period 2010–2016. We, then, employ term frequency-inverse document frequency (TF-IDF) to measure the importance and relevance of weak/strong signal keywords for subsequent business development and utilize Correlation Explanation (CorEx) topic modelling to link weak/strong signals with business concepts of startup companies founded from 2017 until 2020 time period. The ANOVA statistical method is also employed to study the relationship between weak/strong signals and the performance of new businesses. The results show no significant difference between startup companies operating in weak and strong signal related business areas in terms of acquiring venture capital funding and estimated revenue range. Moreover, we examine industry and technology profiles of startup companies by using word co-occurrence analysis and study in which sub-sectors they bring digital technologies across selected European regions. Finally, we discuss the implications of the study for strategic planning and investment policy at the firm and regional levels.