Hacker News Reader

Comment by random42 | Hacker News Reader

Wow, If it works well, it is a boon for people who are not natural cooks, who want to make something homemade, with the ingredients at home.

(I suppose/believe/hope this is a significant market size.)

I would like to know more about the algorithm/science, behind the pairings (and its degree).

View on HN

I was curious about this as well, as it looked to me like a ripoff of "The Flavor Bible", by Page and Dornenburg (Same method of using bold all caps, then bold, etc for showing the important ingredients). However, from the book link, apparently: The data for this book was obtained by programitally[SIC] examining just under a million publicly available recipes.

That method of acquiring the data shows in the less useful results. For example, with basil, it includes in the top pairings SALT, GARLIC, ONIONS, BLACK PEPPER. Which, of course, would be found in pretty much any recipe involving basil. Yet, it misses the more classic pairings like mozzarella cheese, eggs, lamb, pizza, etc.

My recommendation -- buy "The Flavor Bible".

I'm not sure, when talking about what ingredients go well together, you can include "pizza" on the list :D

Thanks! I have indexed about a million recipes with associated ingredients. When someone does a search it retrieves 500 recipes with associated ingredients, using the same algorithm I use for my ingredient search engine. It then counts the ingredient occurrence and factors in ingredient rarity.

I am not sure about the degree of the algorithm for the initial search(using sphinx and its fast enough for me not to care). I believe my algorithm is at worst n^2 because I have to sort the ingredients.

What cook doesn't know which ingredients go well together but is able to make a good dish from just a list of possible ingredients?

Reply on news.ycombinator.com