Aligning Food Ingredients with Multiple Semantic Resources
摘要
To address the lack of integrated resources that combine nutritional, chemical, and semantic information, we have integrated the Recipe1M+ dataset with the USDA National Nutrient Database, FooDB, and the FoodOn ontology. This integration brings significant advantages by enriching the dataset with comprehensive information across multiple domains. By linking Recipe1M+ ingredients to the USDA database, we provide detailed nutritional profiles that enable a comprehensive analysis of recipes, supporting researchers and dietitians in developing personalized nutrition plans based on accurate nutrient data. Furthermore, incorporating FooDB enhances the dataset with in-depth chemical compositions and health effects of food constituents, facilitating research on functional foods and their role in disease prevention and health promotion. Mapping ingredients to FoodOn further expands the dataset’s semantic context, encompassing food products, production, agriculture, and environmental impacts. This integration promotes interdisciplinary research and the creation of comprehensive knowledge graphs, bridging gaps between nutrition, food science, agriculture, and health domains. Our methodology has been validated, demonstrating significantly better results with a precision of 76.25%. Adhering to the FAIR principles, we ensure the data is findable, accessible, interoperable, and reusable. The enriched Recipe1M+ dataset supports advanced applications, such as predictive modeling of dietary impacts, food recommendation systems, and sustainable food system studies, making it a valuable resource for the research community and beyond (The data and code are publicly available at the following link ).