An Open Data Standard For Food

December 18th, 2012

I came across this article by Stacey Higginbotham for GigaOm while doing research for Basil:

An open data standard for food has emerged on the web. With such a tool, restaurants, food apps, grocery stores, the government and other interested parties can tell that arugula is also called rocket salad, no matter where on the web it occurs or what a restaurant menu or recipe app calls it. Right now, that’s an impossible task, which leads to inefficiencies in both consumer-facing apps and the supply chains of restaurants and grocery stores.

A group of folks concerned about sustainable foods have built the seeds of an open food database hosted on Heroku, with the code pertaining to it located at Github.

Really, really cool idea, and something we absolutely need more of. Theoretically, this sort of thing would allow Basil to do a lot of very powerful things. For example, it could have much smarter tagging; rather than just tag recipes with ingredients it uses that happen to be in a built-in list of ingredients or user-added ones, Basil could use the service’s list of ingredients, so you’d get a much fuller tagging system. But it could also be more intelligent about it; if one recipe says it uses “coriander” and another recipe says it uses “cilantro,” Basil could use this service to see oh, they’re the same thing, and to tag both recipes with “cilantro.” Or, if this service ends up providing translations of food names into different languages, Basil could be language-independent: whatever language you save a recipe in, it would display the tags for it in the user’s native language. That’s awesome.

There is huge potential here to do something incredible, and it shows the potential for what open and linked data could do. Imagine if we then take this data-set and link it up to a data-set which provides nutritional information for foods. It would then be quite simple to create a rough estimate of the nutritional content for any recipe, even if the information isn’t provided. From there, we could link up to another data-set which provides the user’s health data (say, Fitbit’s API), and from there, to a service which tells you how many calories, carbohydrates, et cetera someone with that health profile should have each day. All of the sudden, we have a very, very concrete way to recommend recipes to people that meet their health needs. And if they tell Basil that they cooked that recipe today, Basil could update their Fitbit account.

Think about how big a deal that is. We have all of this data already—we just need to unlock it. Open data has the potential to be even more important than the web browser and hypertext.