Tesia Šker, Jože M. Rožanec, Gregor Leban and Dunja Mladenić

ABSTRACT
Strategic foresight helps organizations anticipate future chal-
lenges and opportunities, allowing them to handle uncertainty
better. While strategic foresight is becoming more widely adopted
across organizations, the process still heavily relies on expert
knowledge, and little of it has been automated through artificial
intelligence. In this research, we explore how media news events
can be analyzed to forecast event types that will take place in
the near future. In particular, we consider it a supervised ma-
chine learning problem with a well-defined set of event types and
leverage graph representation of the media news events to create
graph embeddings, train a classifier, and predict event types that
will likely occur one day ahead. We validated our approach on a
real-world dataset of an American multinational conglomerate
operating in industry, worker safety, healthcare, and consumer
goods.