It’s exciting times when you can crowdsource traffic information (Waze) or learn about the popularity of a restaurant (Yelp), or determine if a movie is good or not (Rotten Tomatoes) with just the click of a button. As this information becomes more easily accessed, developers and users are getting more savvy creating and consuming apps that solve needs that have never been solved before.
With innovation all around us, people are looking for tools that will help them in many aspects of their lives. More specifically, people are looking for information that is relevant, timely, and personalized. It is fantastic to see developer communities innovating the solutions of the future.
What about Food Information and Tools?
While innovation is happening in many verticals, personalized and relevant food information is scattered and difficult to find. As a result, innovation in food-tech has been slower to progress than other verticals. Here are some contributors to the slower innovation in the food-tech arena:
- The food industry is slow to change and generally risk averse
- Data is proprietary, not always comprehensive, and fragmented
- Data is very expensive
- Data is not in a structured format so it’s difficult to aggregate
- Government data on food and nutrition is limited
Food, on the other hand, remains one of the most vital, personal, and perplexing parts of our daily lives.
About the Yummly API
Yummly is excited to be changing this paradigm. After several years of building the most sophisticated recipe technology available, we are excited to officially release the Yummly Recipe API to the public. In short, Yummly has indexed over a million recipes from across the Web, built a recipe ontology based on algorithmic learning, and has created the largest, most structured and searchable data set available for recipes in the world.
Over the last year we have been quietly testing with a community of over 600 developers and are extremely excited about the future of the Yummly technology and solutions that can be created leveraging the API. Companies (DuckDuckGo, Mindspark, PushPins, etc.) and Universities (Stanford, Harvard, Berkeley, etc.) are using the Yummly API to power everything from television apps to mobile apps… from wine pairing apps to allergy apps.
To better understand what platforms developers have been utilizing with the Yummly API, we ran a survey to our developer community. Here was the breakdown by platform.
It’s all about the Ingredients and the Prep Steps…
The Yummly API has been developed and is maintained by an Engineering support team of eight engineers and an API product specialist. The team has extensive experience writing and implementing API products. Yummly’s recipe and ingredient metadata is created and validated via a set of trained classifiers and sophisticated algorithms.
The team seems to be heads down today…
Separately, to keep things fresh and relevant, recipes, ingredients, and all of their corresponding metadata are constantly being checked and updated by an analytics and data team. This team consists of 5 people, including machine learning specialists, a chief analyst, a computational linguist, and a semantics specialist.
The Yummly Recipe API supports Yummly.com and all of Yummly’s properties.
What’s New and What’s Coming for API partners?
- Recipe content - Since the beginning of Q1, we’ve increased the number of recipe sources by 60%, and we’re continuing to add new sources and recipes every day.
- Search Improvements -This month we released an improved search algorithm that exhibits our understanding of semantic relevance in recipe search. The new algorithm allows users to query all of their dietary and recipe needs; users can modify their search with one or a combination of restricted diets, allergies, cooking techniques, source names, cuisines and courses.
- Machine Learning and Information Retrieval Applications - One of the most exciting parts of our new search algorithm is that our search results continue to evolve as we collect more user data. Clicks and yums on Yummly properties drive our relevance model, so the more users click on a given recipe in a set of results, the higher it will appear in the results. At the same time, we try to maintain diversity in the results by balancing relevance with recipe diversity.
- Recipe Data and Knowledge - Yummly’s Data and Analytics team is continuously working to improve the company’s deep knowledge of recipe data. Our recipe data is what sets Yummly apart from any other recipe technology. First, we collect a number of attributes and data points from all of our recipes, including nutrition, course and cuisine information, the techniques used to make the recipe, and more. Then, we utilize our semantic ontology and machine learning classifiers to extract as much data as possible from every recipe we index. This way, we are constantly improving the way we meet our users’ needs, whether it’s cooking with a gluten allergy, a vegan diet, or a preference for spicy foods. This work directly benefits our API partners as the quality of what we deliver over the API is constantly improving.
- Coming Soon - We are working to roll out some awesome additional features, including personalized recipe recommendations and the ability to sort by up-to-the-minute trending recipes, most popular recipes on the Web right now, and newest recipes.
A Shout-out to our API Partners
We would especially like to thank our API partners who have given us inspiration and guidance to take our API to the next level. We look forward to the amazing apps you will be creating leveraging our technology. Cheers to innovation!
Lastly, Visit our New Developer Portal
We have launched a brand new developer portal with new plans tailored to the needs of everyone from students to enterprise. Check it out at developer.yummly.com.
Let us know your thoughts! We’d love to hear from you.