Studying the content and impact of news articles has been a recurring interest in economics, finance, psychology, and political and media literature over the last 20 years. Most of these offerings focus on specific qualities or outcomes related to their textual data, which limits their applicability and scope. Instead, we use novel datasets that adopt a more holistic approach to data gathering and text mining, allowing texts to speak for themselves without shackling them with presupposed goals or biases. Our data consists of networks of nodes representing key performance indicators of companies, industries, countries, and events. These nodes are linked by edges weighted by the number of times the concepts were connected in media articles between January 2018 and January 2022. We study these networks through the lens of graph theory and use modularity-based clustering, in the form of the Leiden algorithm, to group nodes into information-filled communities. We showcase the potential of such data by exploring the evolution of our dynamic networks and their metrics over time, which highlights their ability to tell coherent and concise stories about the world economy.