The early diagnosis of catastrophic events: weak signals detection based on information field monitoring

Abstract

Objective: This article explores the possibility of the early diagnosis of catastrophic events based on weak signal detection and information field monitoring. Data and Methods: Publications in The New York Times and The Guardian are used as a source of data. To analyze the dynamics of the information field, the concept of weak signals is applied. Text mining methods (text preprocessing, frequency analysis, word clouds) are used. Statistical methods for analyzing distributions and testing hypotheses are used to determine anomalies in keyword frequencies. Results: The study makes it possible to confirm the assumption that the information field begins to change some time before the real onset of catastrophic events. The changes in the information field that preceded the two selected catastrophic events – the COVID-19 pandemic and Russia’s military invasion of Ukraine – are analyzed. Statistically significant changes in the information field in both cases were observed long before the start of these events. It was revealed that deviations in the frequency of keyword mentions more than 4σ are anomalous, and can signal impending disasters. The assumption that it is possible to track the state of the information field through open media resources is also confirmed. Conclusions: The results of the study confirm the possibility of using the analysis of the information field for the early diagnosis of global catastrophes. The creation of a system for monitoring the information field using the methods of machine linguistics can contribute to the timely preparation and minimization of threats to economic security.