You are here:  Home  >  Data Blogs | Information From Enterprise Leaders  >  Current Article

Edamam's Semantic Smarts Help Serve Up Dinner Plans

By   /  May 16, 2012  /  No Comments

Edamam wants to be the one place where all the food knowledge of the world is organized. That’s the goal of co-founder and CEO Victor Penev, who launched the site in April, and recently updated the several hundred major recipe sites in its knowledge base to also include some smaller blog sites that add additional variety.

Semantic technology is helping the company reach its goal. “A big problem is that data about food is very messy,” says Penev. “It’s hard to find something, what you find often contradicts other information of what is good for you and what the calories are. So we set out to solve that problem. We played around with different approaches but settled on using semantic technology.”

The confusion arises in part from the fact that recipe sites themselves usually just hire services to calculate nutritional data. But that may lead to mistakes when calculations aren’t undertaken with exactitude — substituting white cream for heavy cream nutritional details changes the whole profile of the recipe, he says.

So, what is that right semantic stuff? One piece of it is that, in conjunction with Ontotext, Edamam built a food ontology. An ontology can be the foundation for a lot of things, such as extracting the knowledge of the chemical composition of a particular recipe and thus inferring its flavor and texture. And Edamam means to grow its own to include various datasets such as chemical data (for flavor and texture), geolocation (for local and seasonal recipes), product data (for e-commerce). and more.

But initially, it’s taken the simple approach, with the core of the ontology focused around classifying ingredients, nutrients and food. “We have started with the simplest ontology and focused on the most common use case — mobile recipe search,” he says.

Edamam so far has scraped more than 1 million recipes, using information extraction techniques to draw out their contents, which then go through its natural language processing and machine learning functions before being mapped to the ontology. It’s got some 8,000 foods covered in its knowledge base at this point.

“Most of those have come from the USDA [for the most accurate nutritional information], but we have enhanced it by integrating other databases (open data and private), and manually adding and or correcting foods/ingredients/measures to constantly improve it,” he says. It’s exploring additional sources such as databases of foods that are good for particular diseases, as well as the European equivalent of the USDA. Also, it has a partnership with The French Culinary Institute for its students to help in deepening and expanding its knowledge base.

“It’s powerful semantic technology – NLP and the organization of data and being able to on an on-going way process any recipe out there,” Penev says.

Food’s Future With Edamam

The information it delivers now is just a glimpse of what plans are for the product ultimately to do, such as perhaps offer cooking assistance or even be propagated into smart-store shopping carts to help users determine how to augment their meal plans with what’s on the shelves. Currently its mobile focus covers iPhone and Android apps, though users can access it from their desktops on the web. But as Penev sees it, “people make decisions about what to eat on-the-go.” Users at the supermarket searching its site  can filter information by diet, if they care about high protein or want a recipe under 200 calories, for instance, and in a minute have a recipe and a shopping list for it.

Its search function algorithm ranks recipes based on a variety of criteria, not just the pure text of the recipe but the proportion of its ingredients, cooking techniques, and so on. “Our ranking algorithm has a number of qualitative components which help it bring the better quality recipes up on top. We calculate a number of things, such as difficulty, healthiness and so on, which enter into the ranking calculation. The formula is much more complex and involves some human curation, too,” he notes.

Penev believes there’s opportunity in working with major food sites in bringing its technology right to their doorstep, so that recipe searches conducted on their sites are served right up with the best and most comprehensive nutritional data. This Friday gives a taste of that as it launches a widget for food bloggers; the WordPress plug-in lets a blogger display nutritional information next to each one of its recipes. “It is part of our overall strategy to organize food knowledge and make it as widely available to users as possible (be it B2C as with the recipe search, or B2B as with the food blogger widget),” he says, adding that the widget will be available soon for other platforms as well.

And, says Penev, the whole recipe space hasn’t been as well-monetized as it could be on the consumer side. “Everyone goes with the ad model,” he says. “But we think someone can create a transaction model around recipes and cooking.”






About the author

Jennifer Zaino is a New York-based freelance writer specializing in business and technology journalism. She has been an executive editor at leading technology publications, including InformationWeek, where she spearheaded an award-winning news section, and Network Computing, where she helped develop online content strategies including review exclusives and analyst reports. Her freelance credentials include being a regular contributor of original content to The Semantic Web Blog; acting as a contributing writer to RFID Journal; and serving as executive editor at the Smart Architect Smart Enterprise Exchange group. Her work also has appeared in publications and on web sites including EdTech (K-12 and Higher Ed), Ingram Micro Channel Advisor, The CMO Site, and Federal Computer Week.

You might also like...

Evolving Cloud Networks and How to Keep Them Secure

Read More →