Estimating nutrition values for internet recipes

Abstract

To utilise the vast recipe databases on the Internet in intelligent nutritional assistance or recommender systems, accurate nutritional data for recipes is needed. Unfortunately, most recipes have no such data or have data of suspect quality. In this demo we present a system that automatically calculates the nutritional value of recipes sourced from the Internet. This is a challenging problem for several reasons, including lack of formulaic structure in ingredient descriptions, ingredient synonymy, brand names, and unspecific quantities being assigned. Our results show that our system can generate nutritional values within a 10% error bound of human assessors for calorie, protein and carbohydrate values. Based on our findings this is smaller than the bound between multiple human assessors.