Answers are grounded in this paper and cite their sources.
Text mining reveals challenges in entity detection for food mentions, indicating gaps in current food ontologies.
Funding for nutrition research in the United States is less than 5% of the National Institutes of Health budget, so nutrition researchers often turn to published work. This provides an ideal environment for text mining, where entity detection is the task of finding food mentions in text and entity linking connects each food expression to a specific food. For example, the system should harmonize the expressions soybean , soy bean , soya bean and the scientific name Glycine max (L) Merrill along with their plural forms to a single concept soybean. However, the system must not harmonize soybean with soy sprouts because these different forms of soya foods have very different nutritional profiles. Despite the numerous food ontologies available, our work on developing a gold standard for food entities revealed a unique set of challenges that would limit the utility of automated extraction for nutrition researchers.
No takes yet. Share an insight, caveat, or question.
Blake et al. (2025) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: