Cultures around the world have acquired unique culinary practices reflected in traditional recipe compositions. Data-driven analysis has the potential to provide interesting insights into the structure of recipes and organizational principles of cuisines. We provide a curated compilation of 45772 traditional recipes from over 22 regions across the world. Using this resource in conjunction with data of flavor molecules from natural ingredients, we implement data-driven investigations for probing flavor pairing patterns in these recipes. Our analysis reveals non-random recipe compositions characterized with either 'uniform' or 'contrasting' flavor blending and identifies popularity of ingredients as a key contributing factor across all cuisines. Thus we provide a framework for data-driven investigations of culinary patterns in recipes which can be leveraged for applications aimed at food design, generating novel flavor pairings and tweaking recipes for better nutrition and health.
@article{arxiv.1803.04343,
title = {Data-driven investigations of culinary patterns in traditional recipes across the world},
author = {Navjot Singh and Ganesh Bagler},
journal= {arXiv preprint arXiv:1803.04343},
year = {2018}
}
Comments
Accepted for presentation and publication at the 1st edition of the DECOR 2018 @ IEEE International Conference on Data Engineering (ICDE), to be held in Paris, France from 16-19 April 2018