I have rolled out a few specialty recipe search engines. You can now search for vegan, vegetarian, and gluten free recipes in their own specific websites.
I also have a release a simple API that will let developers search the recipe puppy database by ingredient and/or keyword. I look forward to seeing what people do with it.
For a few months now I have been working on an ingredient database. I am compiling a mixture of information like seasonality, nutritional information, and wikipedia with things I can infer from my large database of recipes like what ingredients go well with each other and what are some of the popular dishes that use a particular ingredient. You can see the database here.
Recently, I have noticed that any cooking recipe that I find on the internet is either not rated or rated very highly (between 4 and 5). I did a quick check of the data to confirm my suspicions. On 2 of the top cooking websites (allrecipes.com and recipezaar.com) I polled the recipes that had exactly one rating. On allrecipes.com 75% of recipes were rated 4 or 5. On recipezaar.com 91% of the recipes are rated 4 or 5.
This is not good for the end user. Everything they search for is rated somewhere between 4 and 5. However, these websites are not doing anything to solve the problem. Take a look at their rating forms:
Neither guides the user as to how to rate the recipe or what the rating even means. Am I rating the quality of food, the clarity of the instructions, how tasty the image looks, etc? It is not surprising that there is a correlation between recipes having a image and having a higher rating. These web properties don’t want a more even distribution of the ratings. They want to be the best place to find recipes, or at least have that illusion.
So how does the user actually find the best recipe for a particular dish? I think the solution is simple. Make a recipe rating widget/javascript bookmarklet, (this brings a standard recipe rating system to all recipe websites) which has a rating system that means something. (ie 0-inedible 1-edible 2-good 3-would make again 4-will make regularly 5-one of the best things I have ever eaten) What does everyone think?
Allrecipes.com
| rating | number of ratings | percentage of ratings |
| 1 | 144 | 5.86 |
| 2 | 122 | 4.96 |
| 3 | 345 | 14.03 |
| 4 | 895 | 36.4 |
| 5 | 953 | 38.76 |
Allrecipes.com(recipes that have images)
| rating | number of ratings | percentage of ratings | |
|
|
51 | 1 | 4.53 |
|
|
48 | 2 | 4.26 |
|
|
135 | 3 | 11.99 |
|
|
433 | 4 | 38.45 |
|
|
459 | 5 | 40.76 |
RecipeZaar.com
| rating | number or ratings | percentage of ratings |
| 1 | 403 | 1.16 |
| 2 | 570 | 1.64 |
| 3 | 2158 | 6.2 |
| 4 | 9059 | 26.01 |
| 5 | 22638 | 65 |
RecipeZaar.com(recipes that have images)
| rating | number of ratings | percentage of ratings | |
|
|
1 | 21 | 0.23 |
|
|
2 | 51 | 0.56 |
|
|
3 | 427 | 4.65 |
|
|
4 | 2286 | 24.89 |
|
|
5 | 6401 | 69.68 |
When I go out to eat, I often use restaurant.com to get my meal for cheaper than I would otherwise. I live in New York City where there are hundreds of restaurants with discounts on restaurant.com. I find my self searching for one restaurant after another to see if the restaurant was reviewed well. After repeating this process about 50 times, I got fed up and automated the process.
Now all the restaurants on restaurant.com are sorted by rating in a very simple format. Let me know what you think. I will be adding a feature that lets you sort the restaurant and city names alphabetically if you want.
I am a big fan of the StackOverflow question and answer sites. Now that they have made it possible for anyone to create their own question and answer site with their system, I have created Ask Recipe Labs, a place to ask and answer cooking and food questions.
About two weeks ago I created a new front end to Recipe Puppy. It worked like this:
The feedback on hn and the fact that it made the front page of digg let me know I was on to something. Rather than switch what Recipe Puppy does, I decided to launch this Recipe finding tool as a new project: CookThing
In the last month I have been working on some major updates to Recipe Puppy.
The first upgrade is the ingredient database. The previous version of Recipe Puppy had just under 800 unique ingredients. That number is now over 1,500. In addition I am constantly adding generic categories, so that a search for cheese will find a recipe with any type of cheese.
The next upgrade is the number of different websites Recipe Puppy searches. We went from 8 to 28 websites and added a few hundred thousand recipes(now searching 650k recipes). Additions include nibbledish.com, simplyrecipes.com, myrecipes.com, bigoven.com, whfoods.com, food.sify.com, bestrecipe.com.au, campbellkitchen.com, deliciousdays.com, recipe4living.com, recipe.aol.com, eatingwell.com, jamieoliver.com, cookeatshare.com, bbc.co.uk/food, kraftfoods.com, vegweb.com, recipeland.com, and 101cookbooks.com. Let me know if you have any suggestions for sites to add. I have 30 more in the queue.
The last upgrade was a small ui change. It is now easier to alter your search on the fly. You can specify an ingredient you have, an ingredient you must use, or an ingredient you want to exclude(or don’t have):
In other news, Recipe Puppy is growing significantly. Recipe now sees 4k visitors daily. Let me know what you think of the changes and/or if you have any suggestions.
Here are a few of the ways you can use the recipe puppy search besides using the recipe puppy website.
New products instead of new features
With Recipe Puppy I initially set out to make a recipe search engine that did everything. Besides being a hard time consuming problem, I found that I was adding features that no one was using. I changed my focus and made Recipe Puppy into a ingredient based recipe search engine that would return higher quality recipes before the lower quality ones.
Then I set out to make other single purpose products using the same technology and data I was using for Recipe Puppy. A variation on the same idea, looking deeper into the data, trying to create a community, a series of diet based recipe search engines, and most recently personalized recipes by email were all born.
The advantages of this approach is that each product is more straightforward, and offers a new opportunity to get press and become popular within a niche.