I think being a "search engine" but starting with a very small database of recipes is not gonna help anyone. I tried searching for recipes by ingredient:
- "bok choy" 0 results
- "bok choi" 0 results
- "arugula" 0 results
- "rocket" 1 result (but for oysters rockefeller, which doesn't list rocket as an ingredient)
- "eggplant" 1 result
If the collection is that small, then maybe people should be able to browse, rather than search?
But even then, it doesn't seem actually useful.
The site does let you "search" without entering a query at all, which at least shows me a page with more than one recipe. I filtered by "vegetarian" as a dietary restriction, and saw stuff, at least. I scrolled until I saw a "Pasta Salad" recipe which stated it takes 835 minutes! Curious as to how that could possibly be, I tried to click through ... only to arrive at a page for "Mini Southwestern Corn Pup Muffins with Fiesta Dipping Sauce" covered by a "Welcome Back" sign-in modal. I can see under the modal that one of the ingredients is "hot dogs", so in addition to taking me to a different result than I had clicked on, it also seems to have taken me to something that didn't match my explicit filter criteria. After dismissing the modal I see that the page for this "result" isn't even the recipe; I have to click through to go to allrecipes. I return to my "Munchy" tab and try to go back to the list, and see that on load, it cleared my dietary restriction filter.
It's even worse than I thought. I simply put in their suggested ingredient to search on: rice. Of the four results I clicked on, only one actually used rice. The other three used rice flour or rice wine vinegar. Technically correct but totally misses the spirit of a useful ingredient search.
1. Instead of hosting the images yourself, you're linking directly to Meredith Corp's image servers. They probably won't appreciate that; if you scaled this out enough they'd likely ban your referrer (and that's if copyright doesn't turn out to be a bigger issue first).
2. 199 recipes is too few. I'd say you don't really have anything to demo, or a good dataset to experiment with on your side, until you have tens of thousands.
3. Your search-by-ingredient and search-by-name both do the same thing; they just make a words list and set it as the query for MeiliSearch. Meaning it just fuzzy matches against all the attributes of every document in the database. That's not any more helpful than a generic search engine. I expect a purpose-specific search engine to take advantage of its domain knowledge; e.g. it should:
i. Know ingredient synonyms
ii. Recognize when an ingredient a trivial product of another ingredient (e.g. egg whites are trivial from eggs, but rice vinegar is not trivial from rice).
iii. Other than those two cases, ingredients matches should not be fuzzy. It's no good if baking power matches when all I have is baking soda, or if buttermilk matches when I have butter.
4. The documents in your database are not well curated (which will make #3 more difficult). You're actually leaning pretty heavily on fuzzy matching. For example, in the DB "vegetarian" is misspelled as "vegatarian". And many ingredients have leading whitespace, including newline characters. Keep your DB clean and you'll be able to take advantage of fuzzy matching when appropriate but still retain the ability to do exact matching.
> "rocket" 1 result (but for oysters rockefeller, which doesn't list rocket as an ingredient)
Seems like an inappropriate fuzz filter. Maybe swap it out for one that just does proper stem matching.
The really weird part is that this occurs not only with the regular title search, but also the ingredient search, and none of this recipe's ingredients have a fuzzy match whatsoever. Wat?!
My problem with recipes is not that i cant find them, its that everything has 32 million results and i dont know which ones are good and which ones are crap. The "reviews" on sites never seem to correlate with good vs bad recipes. If I've eaten something before its kinda better but hard to find new dishes to make if I dont also know the recipe is probably good. Anybody have advice on good recipe sources?
Yes I know what you mean. I think there should be a site that realizes that there are many variants of well known dishes, and they should be categorized as that. Very rarely is there an original, true recipe that is the pristine source for the dish. So when I search bearnaise sauce, I should not get a recipe, I should get a general description, kind of like a Wikipedia page. And then below that, all the different recipes, that can be contributed like on a wiki. Each recipe should have a feedback mechanism so the best ones gets upvoted.
Most of the time - the recipes are pretty similar.
It's usually just a difference in spices, oils (or other fats), or emulsifiers. Or vegan / gluten free substitutes.
Nothing beats knowing your tastes and learning how to actually cook - then you can look at a recipe and know which things you'd rather substitute with something else and how and when, etc.
I tried solving this with a search engine that accounted for reviews with some manual curation, but I realized there's actually an extremely limited set of 'good' recipe sites and kept returning the same 10 sites with a long tail of average sites.
You could improve upon this by establishing site authority for a topic (cuisine), but it still seemed pointless to me.
In the short term I'd appreciate some of those good recipe sites as I dont know what is better or worse. Would you mind sharing those 10 or so are good?
Bob Apetit is consistently good for any American or Italian basics. Biscuits, pies, tomato sauce, Omelettes, chili, just about any Italian pasta dishes, anything like that.
They’re also written well and are pretty easy to follow without long preambles or anything.
I have had pretty good luck with the product forums for the thermomix; the recipes are obviously written for that machine but it’s easy to repurpose them.
Always good to see another recipe search engine! Curious how they're doing it, but a lot of things that I search for don't seem to work. It also seems to use a lot of recipe aggregators in the few results I was able to find.
Just searching for pizza doesn't even return pizza recipes. Just pizza dough and Italian nachos (with a picture of a pizza which isn't even anything like the finished product apparently).
Meanwhile I can go to Google or Bing and find actual Pizza recipes with a box at the top dedicated to recipes.
The database is far too small, go and really populate it & come back! this is an awesome idea, when you are ready.
I was confused though, there appears to be no long story involving a grandmother and a rat named terry who used to nibble at the candles in your cabin. ETC! ;-)
Hi all, after hearing all the feedback, I decided to remove the registration wall behind the website. Feel free to use the website as desired. If anyone is interested in contributing their feedback to the product and testing new features, please feel free to register as I will be reaching out to newly registered users
This is exactly what I need....but i searched by ingredients with 'chicken breast' and 'jalapeno' and it returned 0 results. Then i searched 'chicken' and 'jalapeno' and it returned 1 result. Seems undercooked (pun very much intended)
Hey jnsie, sorry for the poor experience. I'm still working on the product and it definitely is in development mode. Although this sounds unintended, this is currently how the recipe search algorithm works. Because the recipe does not include the ingredient "chicken breast", no results come up. However, multiple recipes include the ingredient "chicken", which lead to you getting a result. I'm working on developing some kind of smart detection algorithm that will ensure you're getting more results and a more accurate result page.
You are making a recipe search app without actually paying attention to the information retrieval part of it, which is arguably the most important thing!
Using any sort of backend that tokenizes (and why not, lowercases and stems) and does the searching for you is what you want as a first step. ES, or any of the newer fast search engines that aren't that concerned about relevance would be good first steps, then depending on how much you could grow, you're going to have quite tricky but enjoyable journey to optimizing for relevance!
Oh hey I built a similar recipe search site with 2 Million recipes a couple of months ago, where you can search by recipe name or filter by ingredients:
The search does not recognize what are individual components nor what should be in the title of a recipe. If I search for "white bread," I get recipes that don't include white bread as an ingredient and are not for white bread. I get mussel recipes that include "white wine" and say to dip bread in the broth. It's ultimately not a useful search.
The product is still in development phase right now. I am collecting emails so I can reach out to people and get their feedback on the product. I'm considering removing that registration wall if it poses to be too much of a nuisance
Seems like a wonderful product, but I'm with the others: I probably won't be using it if you need my personal information first. It's a dark pattern (not that I'm accusing you of being malicious) to force people into disclosing anything to use your app, so I'd recommend adding a simple feature like bookmarks that signed-in users can get as a bonus. Otherwise, I see a lot of red flags as a developer when someone is so upfront about needing my email.
Looks pretty promising. There seems to be quite a lot of frustration with finding recipes, so I think there is a market for something like this.
If you are looking for ways to improve the relevance of the search results, try to prioritize results that contain the exact search terms in the title, then some of the terms in the title, and then in the description, and then in the ingredients.
It's pretty tricky to get right, but I don't think you are that far off.
I'd also get rid of the email fishing popovers, it's very off-putting.
Just to add my 2 cents. This is great ideea and for sure I can see the use of it. But the email part, that is not somehting I would do for a recipe. Next, the you can add planner and you will have any nutritionist on this site giving you the emails. and they can fine tune the data you have. Ofc "suggest or improve recipe" is a very good ideea :) best of luck.
Keep in mind that people want to make those "basic" ingredients that are often overlooked. Bread, vinegar, sauerkraut, even butter can be made from scratch at home. They're usually more labor and time intensive, and don't follow the "heat thing up and add to other thing" pattern, but the recipes are out there.
Why would I use this when Google returns fantastic recipe options at the top of any recipe-related search results page, including pictures, ratings, and an ingredients summary?
Because all the current recipe sites are cesspools of adverts, unnecessary and distracting write-ups, and obfuscated printability? They are poster children for much of what makes the current web a miserable chore to use.
I really want this to work, I love the filter options that I see (assuming they work) and I'd love to use something like this instead of Google/Bing (where the top-recommendations are all ads or super bloggy).
I've taken to searching food manufacturer and grocery store websites for recipes. Since they're selling you the ingredients there are no ads, although recipes are more limited in scope.
Hey! That's the goal. I will admit the database is small because I'm adding them manually but I'm working on finding a way to scrape recipes automatically. In the meantime, the tradeoff for the better user experience is less recipes
Is there a way to help out with that? I just take screenshots of recipes on my phone because I hate the modern internet. If I could enter in a url and you could scrape the recipe off of that page, that'd be great (for others).
Out of curiosity, is this a hobby-project or something you're going to try to make money off of?
Yeah, thanks for the idea! I'm planning on allowing manual submission of recipes.
I ultimately want Munchy to become a company. I have a lot of other innovations planned for the consumer home kitchen space and would like to see it through! Making this website a successful business is the first step for me to achieve that goal
If you ever want to share feedback or contribute, my email is on my profile so you are free to reach out.
The great thing about this is that I don’t have to read (skip) a fucking preamble before each recipe about why this dish reminds the author of that time they spent a summer backpacking through Europe with their boyfriend. Keep it up.
I need a recipe for the best cookies that don't harden. Every website does the same ingredients and they're always rock hard cookies. I want recipes from actual bakeries, chef's.
I’m no chef, but I live in a dry climate, and for soft cookies I get good results by storing them in sealed containers with a slice of bread immediately after cooling. The moisture in the bread makes it like a humidor for your cookies so they stay soft longer :)
I don't think it's important to find a recipe that uses all of the ingredients. I think it might be better to find recipes that uses fewest additional ones.
- "bok choy" 0 results
- "bok choi" 0 results
- "arugula" 0 results
- "rocket" 1 result (but for oysters rockefeller, which doesn't list rocket as an ingredient)
- "eggplant" 1 result
If the collection is that small, then maybe people should be able to browse, rather than search?
But even then, it doesn't seem actually useful. The site does let you "search" without entering a query at all, which at least shows me a page with more than one recipe. I filtered by "vegetarian" as a dietary restriction, and saw stuff, at least. I scrolled until I saw a "Pasta Salad" recipe which stated it takes 835 minutes! Curious as to how that could possibly be, I tried to click through ... only to arrive at a page for "Mini Southwestern Corn Pup Muffins with Fiesta Dipping Sauce" covered by a "Welcome Back" sign-in modal. I can see under the modal that one of the ingredients is "hot dogs", so in addition to taking me to a different result than I had clicked on, it also seems to have taken me to something that didn't match my explicit filter criteria. After dismissing the modal I see that the page for this "result" isn't even the recipe; I have to click through to go to allrecipes. I return to my "Munchy" tab and try to go back to the list, and see that on load, it cleared my dietary restriction filter.