Answers are grounded in this paper and cite their sources.
Food systems are going through major changes, requiring the authorities to foresee and address rapidly evolving risks in a timely manner. Natural language processing (NLP) is a subfield of artificial intelligence that facilitates the analysis, processing and generation of human language content. Especially considering the recent developments, this subfield holds promise as a valuable toolkit for food safety research and policy. This review aspires to be a reference source for researchers who wish to leverage NLP for food safety research. We introduced modelling approaches in NLP, provided a taxonomy of the studies and described the labelled datasets. Future avenues of research were also discussed. We examined 103 studies published between 2010-2024. The most common purpose was assisting authorities in general monitoring of hazards (n = 34). Text classification was the most frequently applied task (n = 42). Social media was the most prominent data source (n = 28). Limited sharing of the labelled datasets (4 out of 43 datasets) was considered as a significant challenge with serious implications for reproducibility, open science and inter-organizational collaboration. Potential future work includes prioritization of hazards, adoption of a holistic approach in monitoring, creation of data exchange standards and multi-lingual datasets.
No takes yet. Share an insight, caveat, or question.
Özen et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: