Know more

Finding values of components of interest for similar foods in other food composition databases is one way to address incompleteness in the data. However, an additional problem arises because the vocabulary used to name one food in one database is usually different from that used in others. In this work, we focused on LanguaL, a multilingual thesaurus used in major national food composition databases to describe a food by a set of encoded terms called facets, and FoodON, an ontology that aims to collect the open reference vocabulary for food science.

A new method was developed to automatically find similar foods referenced in different food composition databases through a web application called MultiDB Explorer. The method calculates a degree of similarity between foods using the FoodON ontology, the LanguaL faceted description and the English name of the compared foods. The MultiDB Explorer application was able to overcome the missing values in the ANSES-Ciqual food composition table for 3 constituents of interest (iron, vitamin C, vitamin B12) in 76 foods. In all, 91% of the missing values could be determined and 96% of the known values could be complemented with values from the US Department of Agriculture (USDA) database.