Creating a semantic food knowledge base with cooking recipes for a meal recommender system

The LIFANA Nutrition Solution helps elderly people maintaining a healthy BMI as their metabolism is changing with age and their eating habits eventually need to be reconsidered. LIFANA provides a personalized meal plan that helps users to prevent undernutrition or overweight, targeting the total daily calories consumed and proteins. The LIFANA approach is based on meal recommendations instead of food logging to make its usage more convenient, since only deviations from the plan need to be reported. In order to make personalized recommendations, the user is represented in the system by height, gender, weight, taste preferences, diet restrictions, and activity level. Core of the system is an extensive (1000+) recipe database with semantic annotation, linking the ingredients of each recipe to Food Composition databases (FCDBs) to infer their nutrients. Within Europe, the recipes need to be localized, which means not only translated, but also linked to different national FCDBs. The prototype app for iOS and Android is integrated in two different ecosystems: GoLive, a Dutch wearable device, and the Portuguese retailer Continente that provides home delivery services. Currently, LIFANA is evaluated in field trials in The Netherlands, Portugal, and Switzerland. LIFANA is funded by the EU and AAL Programme.