Unsupervised news analysis for enhanced high-frequency food insecurity assessment

This article introduces an artificial intelligence (AI)-based system for forecasting food insecurity in data-limited settings, employing unsupervised neural networks for topic modeling on news data. Unlike traditional methods, our system operates without relying on expert assumptions about food insecurity factors. Through a case study in Somalia, we show that the method can yield competitive performance, even in the absence of traditional food security indicators such as food prices. This system is valuable in supporting expert assessments of food insecurity, unlocking a wealth of untapped information from news outlets, and offering a path toward more frequent and automated food insecurity monitoring for timely crisis intervention.