An exploration of the potential of machine learning tools for media analysis to support sense-making processes in foresight
In view of the many discussions about uncertainty regarding the further development of the coronavirus disease 2019 (COVID-19) pandemic and its effects on the economy and society, we observed that the crisis led to an increased presence of individual researchers and experts making forward-looking statements on the impacts of the COVID-19 pandemic or stating trends in mass media publications. From a strategic foresight research perspective, there is a need to further analyse an increase of future-oriented expert statements in public media in a context of high uncertainty like the impacts of the COVID-19 pandemic and related crises. Given the increasing amount of media texts available for web-based scanning and text analysis, Machine Learning (ML) is a promising approach for text analysis of big data, which also raises high expectations in the field of foresight, particularly in the context of scoping and scanning activities for weak signal detection and text analysis for sense-making processes. In this study, we apply a natural language processing (NLP)-based ML approach to analyse a large corpus of news articles from web sources to explore the potential of applied ML to support sense-making in the field of foresight, specifically for the analysis of future-related statements or predictive statements in media. The results underline the potential of ML approaches as a heuristic tool to support sense-making in foresight processes and research, particularly by pre-structuring large datasets (e.g., news articles around a particular topic of public debate). The ML can provide additional insights for actor analysis associated with a specific topic of public debate from a large data corpus. At the same time, our results show that ML models are limited in their ability to provide solid evidence and that they can also lead to fallacies. Therefore, an ML can only be considered as a heuristic tool supporting specific steps in a sense-making process and development of further research questions, as well as encouraging reflection on the application of ML-based approaches in foresight.